NUREG-CR-5587 Approaches For Age-Dependent PSAs Etc

NUREG/CR-5587
SAIC-92/1137
Approaches for Age-Dependent

Probabilistic Safety Assessments
With Emphasis on Prioritization
and Sensitivity Studies
Preparedby
W. E. Vesely
Science Applications International Corporation
Prepared for
U.S. Nuclear Regulatory Commission
AVAILABILITY NOTICE
Availability of Reference Materials Cited In NRC Publications
Most documents cited In NRC publicatIons will be available from one of the following sources:
1. The NRC Public Document Room. 2120 L Street, NW., Lower Level, Washington, DC 20555
2. The Superintendent of Documents. U.S. Government Printing Office, P.O. Box 37082. Washington.
DC 20013-7082
3. The National Technical Information Service, Springfield, VA 22161

Although the listing that follows represents the majority of documents cited In NRC publications, It Is not
intended to be exhaustive.
Referenced documents available for nspection and copying for a fee from the NRC Pubic Document Room
Include NRC correspondence and internal NRC memoranda: NRC bulletins, circular, information notices,
inspection and nvestigation notices: licensee event reports; vendor reports and correspondence; Comrmis-
slon papers; and applicant and licensee documents and correspondence.
The following documents in the NUREG series are avalable for purchase from the GPO Sales Program:
formal NRC staff and contractor reports, NRC-sponsored conference proceedings, international agreement
reports, grant publications, and NRC booklets and brochures. Also available are regulatory guides, NRC
regulations In the Code of Federal Regulations, and Nuclear Regulatory Commission Issuances.
Documents available from the National Technical Information Service Include NUREG-serles reports and
technical reports prepared by other Federal agencies and reports prepared by the Atomic Energy Convnis-
slon, forerunner agency to the Nuclear Regulatory Commission.
Documents avaliable from pubic and special technical libraries Include all open literature Items, such as
books, Journal articles, and transactions. Federal Register notices, Federal and State legislatIon, and con-
gressional reports can usually be obtained from these lbraries.
Documents such as theses, dissertations, foreign reports and translations, and non-NRC conference pro-
ceedings are available for purchase from the organization sponsoring the publication cited.
Single copies of NRC draft reports are available free, to the extent of supply, upon written request to the
Office of Administration, Distribution and Mail Services Section, U.S. Nuclear Regulatory Commission,
Washington, DC 20555.
Copies of industry codes and standards used In a substantive manner In the NRC regulatory process are
maintained at the NRC Ubrary, 7920 Norfolk Avenue, Bethesda, Maryland. for use by the pubic. Codes and
standards are usually copyrighted and may be purchased from the originating organization or, If they are
American National Standards, from the American National Standards Institute, 1430 Broadway, Now York.
NY 10018.
DISCLAIMER NOTICE
This report was prepared as an account of work sponsored by an agency of the United States Government.
Neither the United States Government nor any agency thereof, or any of their employees, makes any warranty,
expressed or Implied, or assumes any legal liability of responsibility for any third party's use, or the results of
such use, of any Informatlon, apparatus, product or process disclosed In this report or represents that Its use
by such third party would not Infringe privately owned rights.
NUREG/CR-5587
SAIC-92/1137
RG, R1, R9
Approaches for Age-Dependent

Probabilistic Safety Assessments
With Emphasis on Prioritization
and Sensitivity Studies
Manuscript Completed: May 1992

Date Published: August 1992
Prepared by
W. E. Vesely

655 Metro Place South, Suite 745
Dublin, OH 43017
Prepared for
Division of Engineering
Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission
Washington, DC 20555
NRC FIN L1072
ABSTRACT
Approaches are described for incorporating component aging reliability models into a
probabilistic safety assessment (PSA), or probabilistic risk assessment (PRA), of a
nuclear power plant. These approaches and procedures are described from a technical
standpoint and are not to be interpreted as having any regulatory implications.
Component aging failure rate models and test and maintenance aging control models are
presented for utilization. Different approaches for carrying out the aging evaluations are
given. Demonstrations are given involving prioritizing aging contributors, evaluating
maintenance effectiveness, carrying out time dependent evaluations, and carrying out
uncertainty and sensitivity analyses of aging effects.
lii
TABLE OF CONTENTS
ABSTRACT............................................................................................................iii
LIST OF FIGURES ....................................................... ix
LIST OF TABLES ....................................................... xi
EXECUTIVE SUMMARY ....................................................... xv
ACKNOWLEDGEMENT...................................................................................... xxvii
1. THE AGE-DEPENDENT PSA VERSUS THE STANDARD PSA............. 1
1.0 Introduction..................................................... 1
1.1 The Difference in the Treatment of Component Failures ............... 4
1.2 The Difference in the Treatment of Test, Maintenance,
and Repair....................................................................................... 6
1.3 Results Obtainable from an APSA ............................................... 9
1.4 Issues Associated with an APSA .............................................. 15
2. COMPONENT RELIABILITY MODELS USED IN AN

AGE-DEPENDENT PSA .................................................. 17
2.0 Introduction. ........ ,......... 17

2.1 Definition of Aging for Reliability and Risk Applications ................ 17
2.2 Aging Failure Rate Models .............................................. 21
2.3 Failure Rate Aging Effects Versus Overall Time Trends .................. 24
2.4 Age Dependent and Tume Dependent Component Reliability
Results...........................................................I................................... 28
2.5 Modeling the Aging Control of a Test, Maintenance, or
Repair Activity .............................................. 31
2.6 The Good as New Restoration Model and The Good as Old
Restoration Model .33
2.7 More Complex Maintenance and Restoration Models .37
2.8 Summary of Models and Data Needed to Quantify Component
Reliability and Unavailability Effects of A gng .42
v
TABLE OF CONTENTS CONTINUED
3. APPROACHES FOR TRANSFORMING A PSA INTO AN AGE-

DEPENDENT PSA .............. . ............................. 45
3.0 Introduction ................................................ 45

3.1 Successive Stepwise Evaluations Using a Standard PSA ................ 46
3.2 Substitution of Aging Models Into a PSA ........................................ 50
3.3 The Risk Importance Approach for Evaluating
Aging Effects ............................................... 52
4. APPLICATIONS OF AN AGE-DEPENDENT PSA ................................. 58
4.0 Introduction ............................................... 58

4.1 Questions To Be Answered in Setting Up an APSA Application ..... 59
4.2 Impact of Available Component Aging Failure Rate Data .............. 59
4.3 Impact of Available Test and Maintenance Information .................. 63
4.4 Impact oEPSA Information Which is Available .............................. 64
4.5 Impact of Results Which Are of Most Interest .................................. 66
5. PRIORITIZATIONS OF AGING CONTRIBUTORS ................................. 67
5.0 Introduction ....... 67

5.1 Application of the Risk Importance Aging Approach ....................... 68
5.2 Selection of Aging Failure Rate Models and Data ........................... 70
5.3 Selection of Test and Models........................................................... 73
5.4 Selection of the Formulas for the Component Aging Impacts .......... 75
5.5 Detailed Prioritizations of and Component Contributors .................. 79
5.6 Grouped Components Prioritizations ............................................... 88
Vi
6. IDENTIFICATION OF RISK-DIRECTED AGING MANAGEMENT 94

STRAT E GIES............................................................................................
6.0 Introduction .94

6.1 Risk-Directed Aging Management Strategies .94
7. TIME DEPENDENT AGING EVALUATIONS ....................................... 104
7.0 Introduction ............................................. 104

7.1 Basic Time Dependent Equations ............................................. 105
7.2 Time Dependent Equations-for a Linear Aging Failure Rate............. 106
7.3 Applications of the Time Dependent, Linear Aging Failure Rate
Equations..................................................................................... 108
8. SENSITIVITY AND UNCERTAINTY EVALUATIONS .......................... 111
8.0 Introduction . ............................................. l

8.1 Uncertainty Analysis of Data Used in an APSA .............................. 111
8.2 Sensitivity Studies of the Effects of Different Aging Rates ............. 115
8.3 Sensitivity Studies of the Effects of Different Test and
Replacement/Repair Policies ............................................. 119
9. CONSIDERATIONS IN USING A PSA TO EVALUATE THE RISK

EFFECTS FROM AGING OF PASSIVE COMPONENTS ........................ 124
9.0 Introduction .................................................. 124
vu
9.1 The Role and Use of PSA to Evaluate the Reliability of Passive 124
Components and Their Risk Impacts.
9.2 Aging Component Reliability Models Required for PSA Aging 126
Evaluations......................................................................................
9.3 The Crack Growth Phenomenon . .127
9.4 The Corrosion Phenomenon . .128
9.5 Stochastic Analysis as a Complement to Deterministic Crack 128
Growth Analysis.............................................................................
10. SUMMARY AND CONCLUSIONS ........................ 130
REFERENCES ........................................................ ............................................... 131

APPENDIX: GENERAL FORMULAS FOR AGING COMPONENT
UNRELIABILITIES AND UNAVAILBILITIES ............................................... 132
viii
LIST OF FIGURES
FIGURE 1. BASIC STEPS IN A PSA................................................................ 2
FIGURE 2. THE BATHTUB CURVE OF FAILURE RATE VERSUS AGE.... 5
FIGURE 3. INCREASE IN CORE DAMAGE FREQUENCY FROM AGING

FOR DIFFERENT TEST AND MAINTENANCE PROGRAMS... 11
FIGURE 4. TIME DEPENDENT CORE DAMAGE FREQUENCY

INCREASE DUE TO AGING........................................................ 13
FIGURE 5. SENSITIVITY ANALYSIS EVALUATING THE AGING

CONTROL OF A GIVEN TEST AND MAINTENANCE
PROGRAM..................................................................................... 14
FIGURE 6. QUESTIONS TO BE ANSWERED IN SETTING UP AN

EVALUATION...................................................................... 60
FIGURE 7. CORE DAMAGE FREQUENCY INCREASE DUE TO AGING:

MINIMAL MAINTENANCE VERSUS RISK-DIRECTED
CONTROL..................................................................................... 102
FIGURE 8. RELATIVE CORE DAMAGE FREQUENCY INCREASE DUE

TO AGING: MINMAAL MAINTENANCE VERSUS
RISK-DIRECTED CONTROL....................................................... 102
FIGURE 9. TOM DEPENDENT CORE DAMAGE FREQUENCY

INCREASE AC DUE TO AGING................................................ 109
FIGURE 10. UNCERTAINTY ANALYSIS OF AGING EFFECTS: PLANT A 114
ix
LIST OF FIGURES CONTINUED
FIGURE 11. CORE DAMAGE FREQUENCY INCREASE FOR A

PLAUSIBLE RANGE OF AGING REPLACEMENT AT
FAILURE: MONTHLY TESTING ............................................. 120

PLAUSIBLE RANGE OF AGING: REPLACEMENT AT
FAILURE: TWO DIFFERENT TEST INTERVALS ...................... 121

PLAUSIBLE RANGE OF AGING: REPLACEMENT AT
FAILURE VERSUS REPAIR AT FAILURE: MONTHLY
TESTING....................................................................................... 122
x
LIST OF TABLES
TABLE 1. THE DIFFERENCES IN COMPONENT RELIABILITY

TREATMENTS FOR A PSA AND AN APSA............................. 8
TABLE 2. PRIORITIZED COMPONENT AGING CONTRIBUTORS......... 10
TABLE 3. ISSUES ASSOCIATED WITH AN APSA.................................... 16
TABLE 4. QUESTIONS AND ALTERNATIVE ANSWERS TO

EVALUATE THE AGING RELIABILiTY EFFECTS OF AN
ACTI ITY................................................................................... 34
TABLE 5. THE SUCCESSIVE STEPWISE APPROACH FOR PSA

AGING EVALUATIONS............................................................. 49
TABLE 6. THE SUBSTITUTION APPROACH FOR PSA AGING

EVALUATIONS.......................................................................... 53
TABLE 7. THE RISK IMPORTANCE APPROACH FOR PSA AGING 57

EVALUATIONS..........................................................................
TABLE 8. TIRGALEX AGING RATES USED FOR ACTIVE

COMPONENTS ................................................ ........................... 72
TABLE 9. LOW SENSITIVITY AGING RATES USED FOR

DOMNINANT CONTRIBUTORS................................................ 74
TABLE 10. HIGH SENSUVY AGING RATES USED FOR

DOMNINANT CONTRIBUTORS ............................................... 74
TABLE 11. TIRGALEX SURVEILLANCE TEST EF ICIENCIES ............... 76
xi
LIST OF TABLES CONTINUED
TABLE 12A. CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS PLANT A: BASE CASE, SINGLE
CONTRIBUTIONS....................................................................... 80
TABLE 12B. CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS PLANT A: BASE CASE, DOUBLE
CONTRIBUTIONS...................................................................... 81
TABLE 13A: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS PLANT A: UPPER BOUND CASE,
SINGLE CONTRIBUTIONS........................................................ 82
TABLE 13B: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS PLANT A: UPPER BOUND CASE,
DOUBLE CONTRIBUTIONS...................................................... 83
TABLE 14A: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS PLANT A: LOWER BOUND
CASE, SINGLE CONTRIBUTIONS............................................ 84
TABLE 14B: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS PLANT A: LOWER BOUND CASE,
DOUBLE CONTRIBUTIONS...................................................... 85
TABLE 15. COMPONENT IDENTIFIERS ..................... - . 87
TABLE 16. COMPONENTS PRIORITIZED BY ORDER OF MAGNITUDE

CONTRIBUTION: BASE CASE................................................. 89

CONTRIBUTION: UPPER BOUND CASE............................... 90
xli
I U.S. NUCLEAR REGULATORY. COMMISSI ON DATE: 08/25/93
PACKING SLIP
USNRC ORDER NUMB: A65617-00006

Jo VORA ORDER TYPE: 2
217B NLS
DOCUMENT NO. QTY a COST TOTAL COST COMMENTS
NUREGCR5587, -5
TOTAL DOCUMENTS ORDERED 5

TOTAL DOCUMENTS ENCLOSED 5
LIST OF TABLES CONTINUED

CONTRIBUTION: LOWER BOUND CASE............................. 91
TABLE 19. RELATIVE ORDERING OF THE COMPONENT

CONTRIBUTORS: COMBINATION OF THE THREE CASES. 92
TABLE 20. ABSOLUTE ORDERING OF THE COMPONENT

CONTRIBUTORS: COMBINATION OF THE CASES............... 93
TABLE 21A. CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS PLANT A: BASE CASE, SINGLE
CONTRIBUTIONS...................................................................... 95
TABLE 21B. CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

COMPONENTS FOR PLANT A: BASE CASE, DOUBLE
CON BUTIO NS ..................................................................... 96
TABLE 22A. CORE DAMAGE FREQUENCY INCREASE FROM AGING

ACTIVE COMPONENTS PLANT A: FIRST
RISK-DIRECTED STRATEGY, SINGLE CONTRIBUTIONS.... 98
TABLE 22B. CORE DAMAGE FREQUENCY INCREASE FROM AGING

ACTIVE COMPONENTS PLANT A: FIRST
RISK-DIRECTED STRATEGY, DOUBLE CONTRIBUTIONS.. 99
TABLE 23A. CORE DAMAGE FREQUENCY INCREASE FROM AGING

ACTIVE COMPONENTS PLANT A: SECOND
RISK-DIRECTED STRATEGY, SINGLE CONTRIBUTIONS... 100
TABLE 23B. CORE DAMAGE FREQUENCY INCREASE FROM AGING

ACTIVE COMPONENTS PLANT A: SECOND
101
RISK-DIRECTED STRATEGY, DOUBLE CONTRIBUTIONS..
TABLE 24. RELATIVE AGING FAILURE RATE CATEGORIES................ 118
xiii
EXECUTIVE SUMMARY
Overview
Explicit consideration of the risk effects of aging has been an important feature of the
Nuclear Plant Aging Research (NPAR) program being conducted by the Office of
Research of the Nuclear Regulatory Commission. By explicitly considering the risk
effects of aging, aging contributors can be prioritized according to their risk importance.
The aging contributors include active and passive components which are susceptible to
aging and include the aging mechanisms and stressors which can cause component aging.
By prioritizing aging contributors according to their risk effects, aging research and
aging activities can thereby be focused on the risk important areas.
In addition to risk prioritizing the aging contributors, explicit consideration of the risk
effects of aging allows aging management schemes to be explicitly evaluated for their
risk effectiveness in controlling aging. Scheduled maintenance, corrective maintenance,
operational testing, and condition monitoring can all be evaluated for their risk
effectiveness in specific situations. Risk effective combinations of activities and risk
effective schedules can thereby be identified. Furthermore, by focusing on the risk
importance contributors, the risk effective aging management strategies can be made cost
effective.
Finally, explicit consideration of the risk effects of aging allows component failure data
to be evaluated for aging effects and associated risk implications. Aging of single
components and simultaneous aging of multiple components exhibited in data can be
evaluated for their risk effects. Because the risk effects of aging are not necessarily
additive, the risk effects of aging of a single component can be insignificant but the same
aging exhibited by several components can be extremely risk significant. The risk
significant aging effects exhibited in data are of high priority and their causes need to be
evaluated to assure that research programs and aging management programs focus on
these causes.
xv
Age-Dependent Probabilistic Safety Assessments
Because of the usefulness and importance of explicitly considering risk effects of aging,
the NPAR program supported the development of a methodology for age-dependent
probabilistic risk assessments (PRAs) and age dependent probabilistic safety assessments
(PSAs). This report describes the procedures which have been developed for
transforming a PRA or a PSA to an age-dependent evaluation. A probabilistic risk
assessment, or PRA for short, has become a standard approach for modeling and
quantifying accidents and their consequences which can occur at a nuclear power plant.
A probabilistic safety assessment, or PSA for short, is a PRA which focuses on accidents
leading to core damage and which quantifies the core damage frequency, but does not
extend the evaluations to quantify the associated consequences, such as the expected
manrem released. The age-dependent approaches which have been developed to date in
the NPAR program have focused on age-dependent PSAs, however the general
approaches are also applicable to age-dependent PRAs.
There are three basic differences between a standard PSA and an age-dependent PSA, or
an APSA as we will term the age-dependent PSA. These three differences, which also
apply to a PRA, are listed below:
1. An APSA explicitly models aging effects in component failure rates, which generally
cause the failure rates to increase with age, while a standard PSA assumes component
failure rates are constant.
2. An APSA explicitly models the effects of test and maintenances in controlling the
aging of components while a standard PSA does not.
3. An APSA explicitly calculates the aging effects and age dependence on the core
damage frequency and system unavailabilities, while a standard PSA does not and
instead calculates constant values for the core damage frequency and system
unavailabilidies.
The Uses of Age-Dependent PSAs
Because an age dependent PSA, or APSA, explicitly models and evaluates aging effects
on the core damage frequency and system unavailabilities, an APSA can be used in
xvi
various ways to evaluate the risk effects of aging.' Three of the principal uses are listed
below:
1. Aging effects on passive and active components can be explicitly prioritized with
regard to their resulting impacts on core damage frequency and risk. Aging control
activities can thus focus on the risk important aging contributors.
2. The risk effectiveness of given aging management programs, including specific test
and maintenance strategies, can be explicitly evaluated and risk effective strategies can
be identified.
3. Failure data and other experience data at a plant can be input to an APSA to monitor
aging effects on risk to provide feedback to the aging management program being
conducted at the plant.
Issues Associated with Age-Dependent PSAs
The issues associated with an APSA generally involve a lack of data to accurately
determine the aging component failure rates and a lack of test and maintenance aging
control information. At the present time, age dependent component failure rates and test
and maintenance aging control parameters are indeed very sparse. Because of the present
lack of data, the work in the NPAR program has focused on developing approaches
using APSAs which do not necessarily require accurate data. The emphasis of the work
has consequently been on prioritization of aging contributors and sensitivity studies of
aging effects and of aging management program effectiveness which do not necessarily
require precise data or information. This report describes the specific procedures which
have been developed for carrying out APSA evaluations with this emphasis.
Approaches for Age-Dependent PSAs
An earlier NUREG, NUREG/CR-5510 (1), was published in June 1990 and described the
basic methodology that was developed in NPAR for age dependent PSAs or APSAs.
However, detailed procedures for utilizing APSAs for specific evaluations were not
"Risk" is used in a general context here and includes the core damage ferquency, as well as public
health risks.
xvii
covered. This report describes these detailed approaches, and specific evaluations that
can be carried out using presently available information. The chapters of the report are
summarized below:
1. The Age-Dependent PSA Versus the StandardPSA
The differences between an APSA and a standard PSA, which were summarized in the
previous discussions, are described in some detail in this first chapter. The table on the
next page highlights these differences.
2. ComponentReliability Models Used in an Age-Dependent PSA
This chapter describes specific aging component failure rate models and specific test and
maintenance aging control models which can be used in an APSA. The aging failure rate
models which are described cover the spectrum of aging behaviors which are likely to be
exhibited by nuclear plant components and include the Weibull failure rate model, the
linear failure rate model, and the exponential failure rate modeL The test and
maintenance models which are presented cover the spectrum of aging renewal activities
which can occur at a plant and include corrective maintenance models, preventive
maintenance models, and piecepart maintenance models. 1
3. Approachesfor Transforminga PSA into an Age-Dependent PSA
This chapter describes three approaches that can be used to incorporate aging evaluations
into a standard PSA to transform it to an APSA. Procedures for each approach are given
along with the strengths and limitations of the approach. One of the approaches is the
approach described in NUREG/CR-5510, which provides detailed aging contributor
prioritizations, including multiple component aging effects. In this prioritization
approach, appropriate risk importance coefficients are extracted from the standard PSA
and are combined with component aging models. Hence, the approach is efficient since
existing PSAs can be directly used with minimal requantification required.
4. Applicationsof an Age-Dependent PSA
This chapter describes the different applications of an APSA and the questions that need
to be addressed in setting up any application. The figure on page xvi presents the
specific questions that need to be addressed to determine the specific aging models and
XVIII
THE DIFFERENCES BETWEEN A STANDARD PSA AND AN AGE
DEPENDENT PSA (AN APSA)
PSA APSA
Component Failure Constant Age-Dependent
I Rates Failure Rates Failure Rates
Surveillance Tests Only Affects Component Effects on Component Age
. ~~Upor Down Status Also Modeled
Maintenances Only Component Effects on Component
Downtimes Considered Age Also Modeled
Repairs Does Not Affect Age Effects on Component Age
Also Modeled
Risks Results Calculated Do Not Explicitly Explicitly Accounts

Account for Aging for Aging Effects
Prioritizations of Based on Constant Risk Aging Contributors
Contributions Contributions Explicitly Identified
Maintenance Maintenance Downtimes Benefit of Maintenance in
Effectiveness Evaluations Only Explicitly Evaluated Controlling Aging Also
:______ for Risk Effects
_ Evaluated
xix
QUESTIONS TO BE ANSWERED IN SETTING UP AN
APSA EVALUATION
Available aging failure rate data?

. - ~~~~~~~~~~~~~~~~~~~~~i
I I
plant
specific generic subjective data
I
Available test and maintenance information?
detailed
I I
minimal
information information
I
Available PSA information?
I
event trees and truncated
fault trees minimal
I n cut sets
Results of most interest?
bottom line
I I
pnoritizaton
and sensitivies
numbers
xx
approaches that are most effectively used. As shown in the figure, the four questions that
need to be asked involve the component aging failure rate data which is available, the test
and maintenance information which is available, the standard PSA information which is
available, and the aging results which are of most interest The specific aging models
and approaches which are most effective for each possible answer to the questions are
defined. What is identified from these descriptions is the aging models which are
available for a given application, and the applications which can be carried out, when
there is little to detailed aging data and information.
5. PrioritizationofAging Contributors
This chapter demonstrates the use of an APSA to prioritize aging contributions with
regard to their risk impacts. This application is important since it can provide a basis for
focusing aging management activities. Those aging contributors which have significant
risk impacts are most important and need priority attention. Prioritization procedures are
described along with interpretation of the results. The tables on the next page illustrate a
prioritization which is obtained as a part of a demonstration using a plant specific PSA.
The components in the plant are prioritized in terms of the average increase in core
damage frequency AC which is caused by component aging between replacements of the
components. The average increase AC applies to each year, and the new core damage
frequency for any year is CO + AC where CO is the baseline PSA core damage frequency.
Every component in the PSA is prioritized for its aging contribution to core damage
fiequency and every interaction of aging components is also prioritized for the
interaction contribution.
The two tables on the next page show the top 90% individual component aging
contributors and the top 90% two-component aging interactions for given aging data.
The tables rank the contributors, beginning with the highest aging contributor to the core
damage frequency (cdf). In the top table, the component name as used in the standard
PSA is given along with the importance of the component (I) as calculated from the
standard PSA. The component unavailability q and the increase in component
unavailability Aq due to aging are then given. The aging contribution to the core damage
frequency AC is given in the last column and is the increase in core damage frequency
(above the PSA value) caused by aging of the component The aging contribution AC is
simply the product of the importance I of the contributor and the aging effect on the
contributor Aq which is extremely useful for applications. The table below is similar
xxi
TABLE 1. DOMINANT SINGLE COMPONENT AGING CONTRIBUTORS FOR A SPECIFIC BWR
(ACTIVE COMPONENTS ONLY)
Component Component Aging

Rank Component ID importance I Unavailability Factor AC
q Aq Y/yr)
I ESW-AOV-CC-CCP 9.70E0S 1.OE-03 2.9EO1 2.8E-0S
2 EHV-AOV-CC-CCP 6.34E-0S .OE-03 2.9B-O1 t.8E-OS
3 ESW-AOV-CC-0241B 3.68E-0 1.OE-03 2.9E-01 I.IE-OS
4 ESW-AOV-CC-0241C 3.68E-05 1.OE-03 2.9E-01 1.1E-05
5 EHV-SRV-CC-RV2 2.53E4OS 3.OE-04 2.9E-01 7.OE-06
6 EHV-SRV-CC-RV3 2.5 3E-05 3.OE-04 2.9E-01 7.OE-06
7 DCP-BAT-LP.CCF -2.16E-04 1.08E-03 1.9E-02 4.1E-06
8 Ha-MOV-CC-MV14 SA2E-06 3.OE-03 2.6E-01 1.4E-06
9 HCI-MOV-CC-MV19 5.42E-06 3.OE-03 2.6E01 IAE-06
10 ACP-DGN-FR-EDGC 2.09E-0 1.6E02 3.3E-OQ 7.OE-07
AC = Core damage frequency increase due to aging
TABLE L DOMINANT DOUBLE COMPONENT AGING INTERACTIONS FOR A SPECIFIC BWR

(ACTIVE COMPONENTS ONLY)
Joint Aging Aging

Rank Component ID Component ID Imponance I Factor Factor AC
_ _ql Aq2 O
I ESW-AOV-CC-0241B ESW-AOV-CC-0241C 134E-03 2.9E-01 2.9E-01 1.I-04
2 ACP-DWN-LP-EDGB ESW-AOV-CC-424IC 8.50E-04 3.3E2- 2.9E-01 8.1E-06
3 ACP-DGN-LP-EDGC ESW-AOV-CC-0241B 8.50E-04 3.3E-02 2.9E-01 8. IE-06
4 ACP-DON-LP-EDGC EHV-SRV-CC-RV2 7.69E-04 3.3E-02 2.9E-01 7.1E-06
5 ACP-DGN-LP-EDGB EHV-SRV-CC-RV3 7.69E-04 3.3E-2 2.9E-01 7.1E-06
6 ACP-DGN-FR-EDGC ESW-AOV-CC-0241B 4.79E-04 3.3E-02 2.9E-01 4.6E-06
7 ACP-DGN-FR-EDWB ESW-AOV-CC-0241C 4.79E-04 3.3E-02 2.9E-01 4.6E-06
8 ACP-DWN-FR-EDWC EHV-SRV-CC-RV2 4AOE-04 3.3E-02 2.9E-01 4.12-06
9 ACP-DGN-FR-EDGB EHV-SRV-CC-RV3S 4.40E-04 3.3E-02 2.9E-01 4.1E-06
10 ACP-DGN-FR-EDGB ACP-DGN-FR-EDGC S.34E-04 3.3E-02 3.3E-02 5.9E-07
xxii
except the importance I is the joint importance of the two contributors from the PSA.
The aging contribution AC represents the additional core damage frequency increase
from the simultaneous aging of the components. AC is the product of the joint
importance I and the two aging effects on the components ( Aql and Aq2).
As demonstrated in the tables, it has generally been found that relatively few aging
contributors dominate, even though many components may be aging. The importance of
the contributor (I) as given by the PSA and the aging effect in the contributor (Aq) must
both be properly taken into consideration to obtain the aging contribution to the core
damage frequency. The procedures presented do this proper evaluation. Furthermore,
the aging interactions from the simultaneous aging of multiple components must be
considered since these interactions are nonlinear and are often the dominate aging
contributions, being often larger than the single component aging contributions.
Prioritization of aging contributors does not necessarily require precise data and can be a
powerful tool to focus aging management activities.
6. Identificationof Risk-DirectedAging Management Strategies
Once aging contributions are prioritized, aging management programs can be directed to
the dominant risk contributors. This chapter illustrates how risk-directed aging
management strategies can be defined and can be evaluated. The aging contributors
prioritized in the previous chapter are used to define two alternative risk-directed
maintenance and replacement programs. By carrying out more frequent maintenances
and replacements on the 24 dominant components, it is shown that the aging effects on
core damage frequency are basically controlled so as to be less than the baseline core
damage frequency.
7. Time dependent Aging Evaluations
This chapter describes how detailed time dependent aging results can be obtained from
the PSA using the models and approaches given in the previous chapters. Detailed time
plots are obtained showing the progression of aging impacts on the core damage
frequency and system unavailabilities. The time dependent results show when in a
plant's lifetime aging effects will become significant from a risk standpoint and when
actions need to be taken. Examples are given for a specific plant showing aging effects
having significant impacts on the core damage frequency at a plant age of 15 years
Xxiii
because of insufficient maintenances and replacements on the risk dominant components.
Once the contributors are identified the aging impacts are controllable.
8. Sensitivity and Uncertainty Evaluations
Another important use of an APSA is for sensitivity and uncertainty evaluations of the
risk effects of aging. Even where aging data are sparse, appropriate sensitivity and
uncertainty evaluations can provide useful information on aging sensitivities and on the
risk effectiveness of aging management programs. In this chapter, procedures are
demonstrated for carrying out sensitivity and uncertainty analyses to evaluate plant
sensitivities to aging, to evaluate capabilities of aging management programs, and to
identify aging management strategies which effectively control risks from aging over a
spectrum of plausible aging behaviors.
The figure on the next page illustrates a sensitivity study that is demonstrated to evaluate
two policies: repair at failure versus replace at failure. Under a repair policy, when a
component fails then minimal repairs are made; aged pieceparts are not replaced but are
again made operational. For a replacement policy aged pieceparts are replaced with new
parts at failure. The y-axis (AC) is the average increase in core Oamage frequency over
40 years due to aging, where AC is the increase above the standard PSA core damage
frequency value. The average increase AC again applies to each year. The x-axis is the
plausible range of aging exhibited by the components, expressed as a relative percentage
increase per year in the component failure rate due to aging.
For the sensitivity study, all components were assumed to have the same relative aging.
The same surveillance test intervals were assumed for both policies. The figure shows
that the replacement policy controls the core damage frequency increase due to aging for
a wide range of plausible aging behaviors. The repair policy does not control the risk
effects due to aging, even for relatively small aging behaviors. From the prioritization of
aging contributors, which can be carried out in parallel, the dominant aging contributors
can be identified and replacements only carried out on the important contributors. Thus,
an effective aging management policy can be identified which does not necessarily entail
major costs. Similar sensitivity studies can be carried out for other proposed aging
management programs or maintenance programs.
xxiv
CORE DAMAGE FREQUENCY INCREASE: REPLACEMENT VERSUS REPAIR AT FAILUIRE
1.OE-02
AC 1.OE-03 ........ .
1.OE-04.....
1.OE-046....____________
1-0
.O
E7 . . . ... . .. ..I .. . . .. .. .I . .. .. . . . .. .. .. . . . ... . . . . ... . . . . ..I.. . . . ... . ... .
0%................%....... 50%.........................100%I.........
...... Aging...... Rate.... (%..................in...Failure.......Rate....Per Year)...
Relative....... ...
9. Considerationsin Using a PSA to Evaluate the Risk Effectsfrom Aging of Passive
Components
Finally, this chapter discusses special considerations, beyond those already discussed, for
evaluating the risk effects from aging of passive components. The discussions focus on
more detailed analyses which can be done to estimate passive component aging failure
rates from basic phenomenological models covering crack growth and corrosion. It is
planned that this subject will be covered in greater depth in a subsequent report
Conclusions
The report which has been developed presents detailed models and systematic procedures
for incorporating aging evaluations into a probabilistic safety assessment (PSA) to
explicitly evaluate the risk effects due to aging. The approaches can also be applied to a
probabilistic risk assessment (PRA). The resulting age-dependent PSA (or age
dependent PRA) will allow a spectrum of important applications to be carried out,
including prioritization of aging contributors according to their risk importance,
evaluation of the risk effectiveness of existing aging management programs, and
identification of risk effective aging management policies. Even when aging data are
sparse, applications can be carried out to identify the potentially dominant aging
contributors, the risk sensitive maintenance practices, and the robust programs which can
be carried out to control aging impacts over ranges of plausible aging behaviors that can
exist.
xxvi
ACKNOWLEDGEMENTS
Acknowledgement is given to G.H. Weidenhamer, the NRC project monitor, for

providing competent and practical direction to the work and to LP. Vora of the NRC for
his ideas and inspirational guidance. Acknowledgement is given to P. Kafka,
GeselIschaft fur Reactorsicherheit (GRS) mbh, for his discussions with the author on the
reliability treatments of passive components. Finally, acknowledgement is given to S.M.
Scalzo for her invaluable assistance in preparing this report.
xxvii
1. THE AGE-DEPENDENT PSA VERSUS THE STANDARD PSA
1.0 Introduction
This report describes procedures for explicitly incorporating component aging

evaluations into a probabilistic safety assessment of a nuclear power plant. A
probabilistic safety assessment, or PSA as it is termed, is a probabilistic model of
accidents which can occur at a nuclear plant which can lead to core damage. A PSA
focuses on the evaluation of the core damage frequency as opposed to a probabilistic risk
assessment (PRA) which also evaluates the resulting consequences and health risks from
a core damage event.
Figure 1 is an overview of the basic steps in a PSA. A set of initiating events is first
identified which require safety system responses. For each initiating event, an event tree
is constructed to define the sequences of system responses (success or failure) which can
occur for the given initiating event. The consequence of each sequence in the event tree,
in terms of whether a core damage occurs or not, is determined from plant response
considerations.
For each defined system failure in the event tree, a fault tree is constructed to identify the
basic component failures which can cause the system failure. The fault tree is used to
quantify the system failure probability using component reliability models and
component data. Human errors, test contributions, and maintenance contributions are
included as causes of components being down, in addition to failure causes. The system
fault trees are incorporated into the event trees to identify the component contributors to
the accidents and to quantify the accident sequence frequencies using component data.
The frequencies of the individual accident sequences leading to core damage are finally
summed to provided the core damage frequency. Additional details for carrying out a
PSA or PRA are given in References 1-3, including specific models and formulas which
are used, along with associated computer programs.
In standard PSA evaluations, aging of components is not explicitly included in the

component failure models which are used to quantify the system failure probabilities and
the core damage frequency. When the effect of component aging is to be specifically
evaluated then the standard PSA component failure models need to be modified. Aging
of components can significantly increase the component failure probabilities and can
1
Safety Functions
IE IE
.....................> ..............
Identify and Categorize

Initiating Events (IE's) Develop Functional
Event Trees
..................................................................................................................................................................
IE 61
>..............> ................
Develop System Develop System

Event Trees Fault Trees
.................................................................................................................. ........................................
.................................................................................
Importance
'...................{ CDF
Integrate Fault Trees Determine Core Damage

into Event Trees Frequency and Contibutors
FIGURE 1. BASIC STEPS IN A PSA

significantly increase the core damage frequency, if the aging is not effectively
controlled by tests and maintenances. The next section, Section 1.1, briefly describes the
differences in component failure treatments which are needed when aging is to be
explicitly evaluated using the PSA. This area is described in further detail in Chapter 2.
Standard PSAs also do not explicitly model the effectiveness of tests and maintenances in
controlling aging effects. When the effect of aging is to be explicitly evaluated by the
PSA then the standard PSA models for test and maintenance need to be modified to
explicitly consider their effectiveness in controlling and mitigating aging effects. Section
1.2 briefly describes the differences in test and maintenance treatments that are needed to
evaluate the aging control of test and maintenance activities. Detailed treatments are
again provided in Chapter 2.
When aging effects are explicitly included in the PSA by using component aging failure
models and test and maintenance aging control models then the PSA is transformed to an
age-dependent PSA. As an abbreviation for an age-dependent PSA we shall use the term
"APSA" where the first "A" denotes "Age-Dependent". The results which are obtainable
from an APSA and the applications of an APSA are highlighted in Section 1.3. Finally,
as the last part of Chapter 1, special modeling and evaluation issues associated with an
APSA are highlighted in Section 1.4. Chapter 1 thus provides an overview of the
differences between a PSA and an APSA.
Chapter 2 describes the detailed component models and test and maintenance models
which can be used in an APSA. The focus of Chapter 2 is on modeling the reliability
effects of aging of active components, such as pumps, valves and circuit breakers. The
aging failure rate models in Chapter 2 can also be applied to passive components,
however the aging failure rates may need to be determined using techniques different
from those used for active components. Chapter 3 describes procedures for actually
transforming a PSA to an APSA. Chapter 4 describes different applications of an APSA
and their requirements. Chapter 5 presents example applications of an APSA focused on
prioritizations of aging contributors. Chapter 6 demonstrates how APSA evaluations
can be used toidentify risk-effective aging management strategies. Chapter 7 describes
how time-dependent applications can be carried out using an APSA. Chapter 8
demonstrates how aging sensitivity and uncertainty evaluations can be carried out.
Finally, because the previous chapters focus on active components, Chapter 9 discusses
3
specific considerations for incorporating passive components and their aging impacts into
a PSA. It is planned that a separate report will be issued on the treatment of passive
components in risk evaluations of aging.
1.1 The Difference in the Treatment of Component Failures
The component failure rate is the basic data used in a PSA to determine the failure
probability and the unavailability of a component. Basically, the component failure rate
is the probability of a component failure per unit time given no previous failure. A
standard PSA assumes the component failure rate is constant and is the same value for all
component ages. When aging is explicitly considered then the change in component
failure rate as a function of component age must be considered. An APSA allows the
component failure rate to be a function of age and to change as the component age
increases. The effect of allowing the component failure rate to be age-dependent can
cause significant impacts on the calculated system unavailabilities and the core damage
frequency.
From reliability theory (see References (4-6)) when age dependence is considered then
the component failure rate generally follows a bathtub curve as shown in Figure 2. At an
early component age the failure rate decreases with age as design, manufacturing, and
installation failures are corrected when they are found. This decreasing failure rate
behavior is termed the bum-in period of the failure rate curve. After bum-in, the failure
rate remains constant reflecting steady state failure behavior. This flat portion of the
bathtub is the constant failure rate period of the failure rate curve. Finally, after the
steady state period, the failure rate increases with age, reflecting wear-out and aging
behavior. This is the aging period of the failure rate. The lengths of the bum-in, steady
state, and aging periods of the bath tub curve can vary for different components and for
different operating environments.
A standard PSA assumes a constant failure rate and hence focuses on the steady state
portion of the failure rate curve. An APSA, in addition to the steady state behavior, also
includes the increasing failure rate portion of the failure rate curve and thus can account
4
Bathtub Failure Rate
)(w)
Burn-in Steady State Aging
Agew
FIGURE 2. THE BATHTUB CURVE OF FAILURE RATE VERSUS

AGE
5
for the transition from a constant failure rate to an increasing failure rate.* Chapter 2
presents specific models which can be used to explicitly include aging in the component
failure rate. These models are standard models which are used in the reliability field.
When an APSA explicitly models the increasing component failure rate with age then as
indicated, significant impacts can occur in the calculated risk results. If the test and
maintenance practices do not control the component aging then the component failure
probability and component unavailability can significantly increase with component age
because of the increasing failure rate. This in turn can cause the system unavailability to
increase with age. When the system unavailability increases with age then the core
damage frequency can increase with age. Besides causing unavailabilities to increase,
the increasing component failure rates can cause the accident initiating event frequency
to increase with age, which can also cause the core damage frequency to increase with
age. Whether the increases occur or not, and the sizes of the increases, depends on the
aging control of the test and maintenance practices. The sizes of the unavailability
effects and the size of the core damage frequency effect are explicitly calculated in an
APSA.
1.2 The Difference in the Treatment of Test. Maintenance, and Repair
Since a standard PSA does not explicitly consider the age of the component, the effects
of testing, maintenance, and repair in controlling aging are not evaluated in a PSA. Also,
since a PSA assumes a constant component failure rate which never changes, the effect
of a test, maintenance, or repair activity on the component failure rate is not considered.
The effect of a test and maintenance in controlling the aging of a component is a special
focus of an APSA. The control of aging is evaluated by determining the change in
component age which results from a test or maintenance activity. The effect of a test or
maintenance on the age of the component is important since the component failure rate
depends on the component age. Hence as the component age is modified through testing
and maintenance, the component failure rate is accordingly modified. The failure rate
modifications which occur change the component unavailability and the component
*The burn-in period of the failure rate can also be included, and can be important when there can be
initial design or manufacturing defects in a new component which is used to replace an aged component.
When the component undergoes qualification testing or burn-in testing, as generally carried out for
nuclear plant components, then the burn-in period isgenerally removed.
6
failure probability which in turn change the system unavailability and the core damage
frequency.
Detailed models for the effects of testing, maintenance, or repair on the component age
are used in an APSA. If the component is replaced with a new component then the age
of the component is set back to the age of a new component. If aged pieceparts are not
replaced then the age of the component is not affected and remains the same. Partial
maintenances which only replace specific pieceparts, but not the entire component, will
cause the age of the component to be set back to a partially restored value. A
preventative maintenance, such as lubricating bearings, will not change the age of the
component since piece parts are not replaced, but will slow the aging process of the
component.
Because of the detailed testing and maintenance models which can be used, an APSA can
comprehensively evaluate the effectiveness of a testing and maintenance program in
controlling component aging and its resulting impacts on system unavailabilities and on
the core damage frequency. Not only is the interval important at which the test or
maintenance is performed, but the specific action which is performed is important.
Chapter 2 describes models which are used in an APSA to model the aging control of
different types of test, maintenance and repair activities. Even when minimal
information exists, these models can be useful for evaluating maintenance sensitivities.
It is important to note that because a standard PSA does not model the removal of aging
effects by maintenance, it can not explicitly model any benefits from maintenance. Only
the negative downtime contribution while maintenance is being performed is explicitly
modeled in a PSA. Thus, only the negative effect due to maintenance is explicitly
modeled. An APSA models both the benefits and negative effects of maintenance.
These effects determine the unavailabilities and core damage frequency which result.
Specific formulas for calculating the unavailabilities, and specific approaches to
determine the core damage frequency are given in the subsequent chapters. Table 1
summarizes the differences in test and maintenance treatments and in component failure
treatments, which are together classified as differences in component reliability
treatments, for a PSA and an APSA.
7
TABLE 1. THE DIFFERENCES IN COMPONENT RELIABILITY
TREATMENTS FOR A PSA AND AN APSA
PSA APSA
Failure Rate Model Constant failure Age dependent or time

rates are used. dependent failure rates are
also used.
Test Model A test only determines A test can affect

whether component is up component age and failure
or down. rate as well as determining
whether component is up
or down.
Maintenance Only downtime for Effects of maintenance in

maintenance is considered. correcting degradations
and aging are modeled, as
well as inefficiencies and
downtime.
Repair Model Repair does not affect the After repair, the age and
age or failure rate of failure rate of component
component. are modified with
appropriate models.
8
1.3 Results Obtainable from an APSA
As for a PSA, the core damage frequency, accident sequence frequencies, and safety
system unavailabilities are obtainable from an APSA. For an APSA, however, these
results now explicitly include the contributions from aging and explicitly quantify the
effectiveness of given test and maintenance programs in controlling aging effects.
An APSA can produce detailed prioritizations of the aging contributors. This is perhaps
one of the most important uses of an APSA since it allows one to focus on the dominant
aging contributors for aging control, for data collection and monitoring, and for
additional analyses. Table 2, for example, ranks the top 95% individual aging
component contributors for a given APSA evaluation. Each line entry gives the rank
(column 1), the specific component (column 2), the risk importance of the component
(column 3), the component unavailbility (column 4), the increase in component
unavailability due to aging (column 5), and the core damage frequency contribution from
the component aging (column 6). As can be seen, for this application the top 95%
contributors consists of only 10 components, and hence relatively few components
contribute to the aging impacts on the core damage frequency, a1lowing one to
effectively focus aging analyses and aging control. Details of prioritization evaluations
of aging contributors are given in Chapter 5. Chapter 6 describes how the prioritized
contributors can be used to identify risk-effective, and cost-effective, aging management
programs.
An APSA can calculate the core damage frequency and system unavailabilities as a
function of plant age to explicitly show the dynamic, time dependent aging effects. The
core damage frequency and system unavailabilities averaged over time periods or
averaged over the plant age can also be obtained to show average aging effects. The
increases in the core damage frequency or system unavailabilities due to aging as
compared to the no aging case can furthermore be obtained to highlight aging increases.
Figure 3, taken from NUREG/CR-5510 (7), shows the average increase in core damage
frequency AC that results from aging at a plant when different test and maintenance
programs are carried out. The average increase AC applies to each year and the overall
core damage frequency for any year is CO + AC, where CO is the baseline core damage
frequency from the PSA. The specific test and maintenance programs are not of concern
.9
TABLE 2. PRIORHTIZED COMPONENT AGING CONTRIBUTORS
Component Component Aging

Rank Component ID Importance I Unavailabity Factor AC
q Aq (/-V)
1 ESW-AOV-CC-CCF 9.70E-05 1.OE-03 2.9E-01 2.8E-05
2 EHV-AOV-CC-CCP 6.34E-05 1.OE-03 2.9E-01 1.8E-05
3 ESW-AOV-CC-0241B 3.68E-05 1.0E-03 2.9E-01 1.IE-05
4 ESW-AOV-CC-0241C 3.68E-05 1.0E-03 2.9E-01 l.IE-05
5 EHV-SRV-CC-RV2 2.53E-05 3.0E-04 2.9E-01 7.OE-06
6 EHV-SRV-CC-RV3 2.53E-05 3.QE-04 2.9E-01 7.0E-06
7 DCP-BAT-LF-CCF 2.16E-04 1.08E-03 1.9E-02 4.lE-06
8 HCI-MOV-CC-MV14 5.42E-06 3.0E-03 2.6E-01 1.4E-06
9 HCI-MOV-CC-MV19 5A2E-06 3.0E-03 2.6E-01 1.4E-06
10 ACP-DGN-FR-EDGC 2.09E-05 1.6E-02 3.3E-02 7.0E-07
10
FIGURE 3. INCREASE IN CORE DAMAGE FREQUENCY FROM AGING FOR DIFFERENT
TEST AND MAINTENANCE PROGRAMS
Plant A
1.OE-O1
1.OE-02
1.OE-03 0
Ac
(eyr) 0
1.OE-04
Baseline
_ Core Melt
Frequeny
1.OE-05 (per yr)
o~ ~~~~~~~~~~~~
:
1.OE-061
-1.-0-1- 0-7l ~ -
L=1 8 T=1 L=72 T=1- L=18 L=72 T=6 L=72

Program Characterization
L = Overhaul interval (in months) for all components

T = Surveillance interval (in months) for all components (if intermediate surveillance is performed)
here. Test and maintenance modeling is discussed in Chapter 2. The different points for
a test and maintenance program represent different possible component aging failure
rates. The results indicate the significant differences that different test and maintenance
programs can have in controlling-aging effects.
Figure 4 illustrates the type of time dependent results which are also obtainable from an
APSA. The figure shows the time dependent core damage frequency increase due to
aging for given component aging failure rates and for a given test and maintenance
program. The sudden drops in the core damage frequency in Figure 4 occur when major
components are replaced in the test and maintenance program. Such time dependent
results can show when in a plant's life, aging effects become significant and whether
components are being replaced or being overhauled frequently enough. The core damage
frequency becomes so high in Figure 4 because key components are not replaced
frequently enough, allowing aging effects to build up to relatively high levels. Chapter 7
describes the application of time dependent aging evaluations.
Sensitivity and uncertainty analyses can also be carried out using an APSA to evaluate
the core damage frequency sensitivities and uncertainties associated with aging effects.
Figure 5 shows the results of a sensitivity analysis which evaluates the aging control of a
given test and maintenance program. For the sensitivity analysis, the component aging
failure rates are systematically increased to account for plausible aging which can occur.
The baseline core damage frequency for the plant is 3x1I 5 per year. As can be
observed, the aging control of this particular test and maintenance program is effective
over a plausible range of aging behaviors, limiting the aging increases to be no larger
than the baseline core damage frequency. Chapter 8 provides further details.
In this report we focus on the use of an APSA for prioritizing aging contributors and for
carrying out sensitivity and uncertainty analyses of aging effects. The models and
approaches which are presented however, can also be used to obtain average core damage
frequency results due to aging or time dependent core damage frequency results due to
aging.
12
FIGURE 4. TIME DEPENDENT CORE DAMAGE FREQUENCY INCREASE DUE TO AGING
AC
(per)
0 100 200 300 400 500

Plant Age (Months)
FIGURE 5. SENSITIVITY ANALYSIS EVALUATING THE AGING CONTROL OF A GIVEN
TEST AND MAINTENANCE PROGRAM
MONTHLY TEST
REPLACEMENT AT FAILURE
3.5E-05
AC3.OE-05., /
AC
(Per Year)
1. E 0 . . . . .. . . . .. . . . .. . . .. . . .. . . . . . . . .. . . .. . . . .. . . . .. . . . .. . . .. . . . .. . . . .. . . . .. . . .
1. E 0 . . . .. . . .. . . .. . . ... . .. . . .. . . .. . . .. . . . .. . . .. . . .. . .. . . .. . . .. . . .. . . .. . . .. . .... . .
5.OE-06 _... . ...
O.OE+00 lI I
0% 25% 50% 75% 100%
Aging Rate (o Change in Failure Rate)
1.4 Issues Associated with an APSA
An APSA requires more data and more extended models than a PSA and hence there are
special issues associated with using an APSA. With regard to data, the basic issue is the
present, general sparseness of failure histories for components and for structures.
Because of this sparseness of failure data, aging failure rates which are estimated from
the raw data will generally have large associated uncertainties. In addition to the
imprecision in failure rate values, the shape of the aging failure rate curve will often not
be known. When'there is no plant specific data available, generic data or expert opinions
will need to be used, which will also generally have large uncertainties. These
uncertainties need to be taken into account in using an APSA for a given application.
The most thorough way of treating these uncertainties is to carry out sensitivity studies or
uncertainty analyses using different aging failure rates. Chapter 8 will illustrate how
such studies can be carried out. Even with little precise aging failure rate data, useful
information can be obtained from an APSA regarding important aging contributors and
sensitivities to aging effects, as will be discussed in the subsequent chapters.
At present, there can also be a lack of information on detailed characteristics of specific

tests and maintenances incontrolling aging to allow their precise modeling in an APSA.
Where information is lacking, approximate models may be used to approximate or bound
the effects of the test or maintenance in controlling aging. Examples of approximate or
bounding models are the "good as old" and "good as new" models as termed in reliability
literature. These models are presented in subsequent chapters. More precise test and
maintenance models are also presented, which allow more detailed analyses to be carried
out if data is available, or which allow sensitivity analyses to be carried out. By using
these models, the aging control of given test and maintenance practices can also be
bounded or can be accurately evaluated where information is available. Perhaps as
importantly, the sensitivity of risks to aging effects under given test and maintenance
practices can be evaluated. The risk important tests and maintenances can thereby be
determined to focus further evaluations and program improvements. The subsequent
chapters describe the bases for these applications, as well as for other applications. Table
3 summarizes the issues which are associated with an APSA.
'15
TABLE 3. ISSUES ASSOCIATED WITH AN APSA
Data Bases Little time-history information exists to estimate age

dependent or time dependent failure rates. Uncertainty and
sensitivity studies need to account for this lack of data.
Failure Rate Models Different failure rate models can cause significant
differences in the calculated results in an APSA.
Sensitivity studies can investigate these effects.
Test and Differences in the test and maintenance models can cause
Maintenance Models significant differences in APSA results. Bounding models
and sensitivity studies can help address this issue.
Results Results from an APSA can be uncertain. Relative results

and more qualitative results can be focused on to address
this issue. Sensitivity and uncertainty analyses can also
identify the meaningful conclusions and interpretations.
16
2. COMPONENT RELIABILITY MODELS USED IN AN AGE-DEPENDENT
PSA
2.0 Introduction
As described in the previous chapter, what makes an age-dependent PSA, or APSA,

different from a standard PSA is the component failure models and the test and
maintenance models, which together comprise the component reliability models which
are used.
This section describes the component failure models and test and maintenance models
which can be used for an APSA. These models are necessary to explicitly evaluate aging
effects and their impacts on system unavailabilities and the core damage frequency. To
begin the chapter, the definition of aging for reliability and risk applications is presented.
The difference between the component age and running time is then discussed. With
these basic concepts defined, the modeling of the aging component failure rate is
described, and specific aging failure rate models which can be used in an APSA are
given. Features of these models and ways these models can be applied are described.
Models are then presented for quantifying the effects of testing and maintenance with
regard to controlling aging. Specific models are presented which can be applied to
surveillance tests, corrective maintenances, preventative maintenances, and repairs.
These models quantify the different effects a specific activity has on the component age
and its associated aging failure rate. Models for maintenance and repair of component
pieceparts are also included. Complete aging control, partial aging control, and
inefficiencies in aging control are included in these models.
2.1 Definition of Aging for Reliability and Risk Applications
The definition of aging for reliability and risk applications is directly tied to the behavior
of the component failure rate. From basic reliability theory (References 4-6) the
definition of the component failure rate at a given time is:
The component failure rate at a given = The probability of component failure (1)
time t per unit time at time t given no
previous failure of the component
17
The failure rate of a component at time t is thus basically the rate of failure at time t
when the component has not previously failed. The failure rate always applies to a given
failure mode (e.g., fail to open, fail to close, etc.) and different failure modes will have
different failure rates.
The standard symbol used for the component failure rate is (t). Thus,
X(t) = The component failure rate at a given time t, (2)
For aging evaluations, it is also necessary to express the component failure rate as a
function of the component age. Let w denote the component age. Then,
(w) = The component failure rate at a given (3)

component age w.
The definition of X(w) now is in terms of the component age:
U(w) = The probability per unit age that the (4)

component fails at age w given no
previous failure of the component.
The failure rate at age w is thus the rate of failure at the given age when the component
has not previously failed. The failure rate as a function of age can be a different shape
and form than the failure rate expressed as a function of time if time and age are not the
same.
The component failure rate at a given time can be related to the component failure rate at
a given age. To relate time t and age w, the age of a component at a given time needs to
be known. This relationship can be expressed as w = w(t). Then the failure rate as a
function of the age can be translated to a failure rate expressed as a function of time.
This failure rate relationship can be expressed as X(t)dt = X(w(t))dw(t), where X(t) is the
failure rate function versus time and X(w) is the failure rate function versus age, i.e., X(t)
18
is given by Equation (2) and X(w) by Equation (3), which are generally different
functions.*
For usual applications, a point in given time t is simply the component age w plus some
constant c (which is the component installation time). For this usual case, t = w + c and
t) = X(w(t)) which is what we will assume for our applications.
For reliability and risk applications the definition of aging is directly related to the failure
rate as a function of age X(w). For reliability and risk applications, a component (or
structure) is defined to be aging if the failure rate as a function of component age is
increasing:
Aging occurs in the component if the failure rate X4w) as afunctionof component age is
increasing.
It is important to note that the definition of aging is related to the failure rate as a
function of age and not as a function of time. The failure rate as a function of age may
be increasing, i.e. may be aging, while the failure rate as a function of time may not. For
aging to occur, the failure rate X(w) does not need to continuously increase with age.
There can be periods in which there is no aging, i.e., in which ;(W) does not increase
with age w. Periods in which the failure rate X(w) is increasing, are aging periods. The
figures below indicate different types of aging which can occur with different types of
age dependent failure rates X(w).
Monotonic Aging
Aw(w
-- -- Aging Perod--- w
*For simpler notation, the two different functions, L(w) and A(t), are not shown with different symbols.
Which failure rate is being referenced, i.e., versus time or age, will be clear from the discussions.
19
Threshold Aging
/.
wo Aging Period w
Bathtb Aging
Aging Period
Irregular Aging
Aging Period
20
All the above failure rates exhibit some period or periods of aging. In general, aging
periods at the end of the component age are of most interest and concern because of the
large reliability and risk effects which can result if the aging continues unchecked.
However, aging in any period can cause significant reliability and risk problems.
2.2 Aging Failure Rate Models
Various parametric models exist which can be used to model component failure rate
aging behavior for reliability and risk applications. The models most often used for
aging are the linear aging failure rate model, the Weibull aging failure rate model, and
the exponential aging failure rate model. These models can represent increasingly more
severe failure rate increases with age. Thresholds can be incorporated in any of these
models to represent aging beginning at some nonzero age w.. In general all the standard
parametric models apply to aging periods occurring at the end of the component age with
a possible initial threshold. To model aging periods occurring at earlier ages followed by
some periods of nonaging, the models can be extended by reinitializing the age variable
w or by defining a variable which is a nonlinear function of the age. Equations and
representative graphs for the linear model, Weibull model, and exponential model are
given below for aging occurring after some arbitrary threshold age w0 . Subsequent
chapters will discuss how these failure rate models are applied in an APSA.
LinearAging FailureRate Model
)(w) X.; w< W. (5)
(w) t + a(w-w.); w w (6)
= initial constant failure rate (7)
a = constant aging rate (8)
w. = threshold age after which the failure rate increases (9)
21
Unear Aging Failure Rate
W0
Weibull Aging FailureRate Model
X(W) =ko w< o (10)
A(W)=A°[w ];w>w0 (11)
where
ko = initial constant failure rate (12)
b = Weibull shape parameter (13)
w = threshold age after which the failure rate increases (14)
Other equivalent forms also exist for the Weibull model but the general property is that
the failure rate increases as some power b of the age.
22
Welbuff Aging Failure Rate
b I
bal
b<l
vho w
Exponential Aging FailureRate Model
X(w) X.0 :w<wO (15)
t(w) =X exp (c(w-wO)): w > wo (16)
= initial constant failure rate (17)
c = exponential scale parameter (18)

and
w. = threshold age after which the failure rate increases (19)
23
Exponential Aging Faikire Rate
)()
wO w
2.3 Failure Rate Aging Effects Versus Overall Time Trends
There is sometimes a confusion between failure rate aging effects and an overall time
trend in the component failure behavior. One may, for example, observe no time trends
in the failure behavior as recorded in data and erroneously conclude that there is no
component failure rate aging effects. It is important to understand the difference
between failure rate aging effects and overall time trends in the failure behavior to
properly model aging effects in an APSA.
The overall component failure behavior as a function of time is generally described by

the failure frequency and to understand the difference between failure rate aging effects
and an overall time trend in the failure behavior one must first understand the definition
of the failure frequency:
The component failure frequency = The expected number of component (20)

at time t failures at time t per unit time
Note that the failure frequency is the rate of failure at a given time regardless of whether
the component previously failed or not. The component failure rate as previously
defined by Equations (1) and (4) is the rate of failure from only first failures. The failure
frequency as defined by Equation (20) is the rate of failure from first, second, third, etc.
failures.
24
The failure frequency is sometimes given the symbol f(t):
f(t) = the failure frequency at time t. (21)
The failure frequency can be estimated from failure histories by counting the number of
failures which occur in given intervals of time and dividing by the time intervals.
The component failure rate versus age X(w) can show aging effects while the failure
frequency f(t) versus time may show no overall trends or aging effects. This is because
the failure frequency generally incorporates the effects of test and maintenance and does
not account for any change in component age if the component is replaced or renewed.
The component failure rate versus age on the other hand gives the failure rate behavior
between maintenances and replacements. When a new component is installed the age
starts over but the time doesn't. These differences are important to understand since
confusion on age versus time can result in statements being made that components are
not aging when indeed they are. This confusion can furthermore result in incorrect data
analysis and incorrect aging modeling being carried out.
To more fully understand the difference between failure rate aging effects and overall
time trends in the failure frequency one also needs to consider the impacts of component
renewals. In reliability terms,
A renewal is a restorationof the component age (22)

back to zero, i.e., w = 0.
Also,
At a renewal thefailure rate is set back to its originalvalue at age w = 0 and (23)
the aging behavior is restarted.
For reliability applications a renewal is defined in terms of its effect on the failure mode
associated with the component failure rate. If a given piecepart of the component (e.g., a
valve actuator) is the principal contributor to the failure mode and the part is replaced
with a new part, then the component age for that failure mode is effectively reset to zero.
If several parts of the component contribute to the failure rate for a given failure mode
then all the parts will need to be replaced with new parts to reset the failure rate age back
to zero. The figure below illustrates the failure rate behavior between renewals.
25
Falkire Rate Behavior Versus Time
ti t2 nM t t3 4
1 A times
of renewa
Note that what is plotted is the failure rate versus time. The failure frequency f(t) versus
time follows the same behavior and incorporates the effects of the renewals. As a
function of time there is no overall trend since the failure behavior is cyclic. The
component age repeats at each renewal cycle. The relationship of the component age w
with time t is shown below.
Component Age Versus nme
0
n
n
a
g
5
2 Time
tI .J 3A4 -*iTSld HemnA
The failure rate versus age behavior is the behavior in the first renewal cycle which is
then repeated for each renewal cycle. The failure rate versus age behavior for each
renewal cycle is illustrated below.
26
Failure Rate Behavior Versus Age
Age w W-t,
This difference between the failure rate behavior versus age and the failure frequency
behavior versus time, or failure rate behavior versus time, is quite important The failure
frequency behavior versus time and the failure rate versus time show no overall time
trend since the behavior is cyclic. The failure rate versus age increases with age and
shows strong aging behavior.
Sometimes, failure data is analyzed and the failure frequency verss time or failure rate
versus time is obtained. This is then mistakenly interpreted as being the failure rate
behavior versus age. The failure frequency or failure rate versus time may show no
overall trend with time. This is then mistakenly interpreted as the failure rate versus age
showing no aging effects. However, when the failure rate is correctly determined versus
age, then aging effects may be observed. Various data analysis techniques which are
used to analyze data for aging effects actually analyze for overall time trends in the
failure frequency. It is extremely important that the proper data analysis be carried out to
differentiate aging behaviors versus age and overall trend behaviors versus time. Failure
rates versus age are required for an APSA when test and maintenance effects on the age
are explicitly modeled. This will be further amplified in subsequent sections.
Instead of complete component replacements or renewals, partial renewals can also take
place at a maintenance or repair where the failure rate age is not reset to zero but to some
lower but nonzero value. This can correspond to replacement of only specific pieceparts
or replacement of a part, not with a new part but with a rebuilt part. The figure below
27
illustrates the effects of partial renewal on the component. As can be seen, an overall
time trend in the failure frequency now can be seen.
Faiure Rate Behmaic Versus Time:

Between Paral Renewals
ti t2 nm*t t3 to
2.4 Age Dependent and Time Dependent Component Reliability Results
Age dependent component failure rates are used in an APSA to calculate component
reliability results. The component reliability results are then used to calculate the time
dependent system unavailabilities and the time dependent core damage frequency. The
time dependent system unavailabilities and core damage frequency are calculated with
regard to the plant age which serves as the reference time. Because they are used to
calculate the system unavailabilities and core damage frequency it is important to
understand the component reliability results which are calculated in an APSA.
The component reliability results calculated in an APSA can include the age dependent
component unreliability, the time dependent component unreliability, the age dependent
component unavailability, and the time dependent component unavailability. These basic
quantities define the age-dependent and time-dependent reliability and availability
behavior of the component. Which characteristics are calculated for a particular APSA
depends upon the calculational approach used in the APSA as will be discussed in
Chapters 3.
The age dependent component unreliabilityis the probability of component failure by a

given age:
28
The age dependent component - The probability that the component (24)
unreliability at age w will fail before age w
The age dependent reliabilityis one minus the age dependent unreliability and is the
probability that the component will not fail before age w. The symbol sometimes used
for the age dependent unreliability is U(w). Thus,
U(w) The age dependent component unreliability. (25)
From standard reliability theory (References 4-6),
U(w)=l-exp-J(w')dw3 (26)
where X(w') is the age dependent failure rate.
The time dependent unreliabilityis similar to the age dependent unreliability but is
defined as a function of time:
U(t) = The time dependent unreliability (27)

= The probability that the component will fail (28)
before time t
- 1exp (dt' (29)
where now X(t') is the time dependent failure rate at time t'. As was indicated, the
function X(t') is generally different than the function (w'). These failure rate functions
will only be the same if the time origin (t=O) is the same as the age origin (w=O) and
there is no test, maintenance or replacement activity which acts to set back the
component age.
29
In addition to the component unreliability (or component reliability) the other basic
component characteristic is the component unavailability. The age dependent
unavailabilityis the probability that the component is down at a given age:
The age dependent component = The probability that the component is (30)
unavailability at age w down at age w.
The symbol sometimes used for the age dependent component unavailability is Q(w).
Thus,
Q(w) = The component unavailability at age w. (31)
The age dependent availabilityis one minus the unavailability and is the probability that
the component is up at a given age. The time dependent unavailabilityis the probability
that the component is down at a given time:
The time dependent component = The probability that the component is (32)
unavailability at time t down at time t.
The symbol often used for the time dependent component unavailability is Q(t):
Q(t) = The time dependent component unavailability. (33)
The time dependent availabilityis one minus the time dependent unavailability and is the
probability that the component is up at a given time.
For an APSA the time dependent component unreliability and time dependent component
unavailability need to be calculated to obtain the time dependent system unavailabilities
and time dependent core damage frequency where the reference time is again the plant
age. The plant age is not generally the same as the component age since maintenances on
the component and component replacements can reset the component age. The time
dependent component unreliabilities and unavailabilities are determined from the age
dependent component unreliabilities and unavailabilities by tracking the component age
with time in the APSA.
30
The specific formulas for the above component reliability characteristics depend upon the
specific aging failure rate model and the specific models used for testing, inspection,
maintenance, and repair. Specific testing, inspection, maintenance, and repair models,
and formula implications, are given in the next section. A summary of the formulas for
the component unreliabilities and unavailabilities is also given in the appendix. These
specific formulas can be used to transform the PSA into an APSA. The next chapter,
Chapter 3, describes the approaches which can be used to actually transform the PSA to
an APSA.
2.5 Modeling the Aging Control of a Test. Maintenance, or Repair Activity
When modeling the aging control of a specific test, maintenance, or repair activity for an
APSA the following three questions need to be asked for the given activity:
1. What is the effect of the activity on the operational state of the component?
2. What is the effect of the activity on the age of component?
3. What is the effect of the activity on the aging of the component?
The answers to these questions determine the appropriate aging control model to be used
for the test, maintenance, or repair activity. For many applications, the activity is
performed on a component piecepart instead of the whole component, and the above
questions then apply to the specific piecepart. We consider each of these questions in
further detail.
1. What is the effect of the activity on the operationalstate of the

component?
For the usual modeling in a PSA which also applies to an APSA, the component can
either be up or down, i.e. can be available or unavailable. Hence for an APSA there is
one of two possible answers:
a. The activity changes the state of the component,
or
31
b. The activity does not change the state of the component.
Testing and maintenance activities which bring the component down for the test or
maintenance must have an associated downtime for the activity. The component is
usually assumed to be in an up state after the activity. If inefficiencies are considered
then there is a probability that the activity will leave the component in a down state due
to testing or maintenance equipment failures, procedure problems, or human errors. The
probability then needs to be assigned for each of these possibilities. If the component is
modeled to have more than two states, as for Markov models, then the state or possible
states after the activity needs to be defined with their associated probabilities. For the
applications discussed here, it will generally be assumed that the component is modeled
as having two states, an up state or down state.
2. What is the effect of the activity on the age of the component?
This is a question regarding the renewal effect of the activity with regard to the age of
the component. The renewal effect is with regard to the failure mode of the component
identified in the PSA (and APSA). There is one of three possible answers to this
question:
a. The activity does not renew the component and hence does not change the age
of the component with regard to the failure mode being considered,
b. The activity completely renews the component and sets the age back to zero
for the failure mode being considered,
or
c. The activity partially renews the component and sets the age to some
intermediate value for the failure mode being considered.
These three answers can be shortened to:

a. No renewal,
b. Complete renewal,
or
c. Partial renewal.
32
For a partial renewal answer, the degree of renewal and the new age after the activity
need to be modeled. For a partial renewal, if possible inefficiencies are also modeled,
then the age after the activity can be larger than the age before the activity. The
proceeding sections describe specific maintenance models which can be used to
determine the effect of an activity on the age of the component.
3. What is the effect ofthe acdvity on the aging of the component?
This is different than the previous question and addresses the effect of the activity on the
aging rate of the component after the activity. For example, a preventative maintenance
consisting of lubricating a pump does not generally renew the pump but slows down the
wearing processes on the pump and hence slows down the aging rate of the pump.
There are two possible answers to this question:
a. There is no change in the aging, i.e., no change in the form of the aging
failure rate,
or
b. There is a change in the aging failure rate and the change needs to be defined.
Specific models for determining the effect of an activity on the aging are also described
in the following sections.
Table 4 summarizes the above questions and alternative answers which determine the
appropriate aging control model to be used for the test, maintenance, or repair activity.
In fact, it can be argued that the questions in Table 4 need to be answered to assess the
aging control of any test, maintenance, or repair activity, regardless of whether an ASPA
is done or not. The models which are presented in the following sections progress from
simpler models to more complex models in addressing each of these questions.
2.6 The Good as New Restoration Model and The Good as Old Restoration Model
The two most straightforward models which can be used to quantify the aging control of
a test, maintenance, or repair activity are the good as new restoration model and the good
as old restoration model. These models are described below in terms of the answers they
33
TABLE 4. QUESTIONS AND ALTERNATIVE ANSWERS TO
EVALUATE THE AGING RELIABILITY EFFECTS OF AN
ACTIVITY
QUESTIONS ALTERNATIVE ANSWERS

What is the effect of the activity on the Changes the state of the component.
operational state? Does not change the state of the component.
What is the effect of the activity on the No effect on the age(age remains the same).
component age? Completely renews the component (age set
back to zero).
Partially renews the component (age
determined by a model).
What is the effect of the activity on the No change in aging rate (same aging failure
component aging? rate used).
Aging rate modified (new aging failure rate
determined by a model).
34
give to the questions regarding the effect of the activity on the operational state of the
component, the age of the component, and the aging of the component.
The Good as New Restoration Model
The good as new restoration model provides the following answers with regard to 1) the
operational state of the component after the activity, 2) the effect on the age of the
component, and 3) the effect of the activity on the aging of the component:
1. The component is in an up state after the activity,
2. The activity renews the age back to zero (to "as good as new"),
and
3. There is no change in the aging behavior after the activity (the same aging
failure rate applies with the age reset to zero).
The good as new restoration model is used to model an activity which replaces or
completely renews the component (or component piecepart if pieceparts are being
modeled). If a component piecepart is the dominant contributor to the component failure
mode being considered then renewal of the piecepart will basically result in renewal of
the component. The good as new restoration model or good as new model for short, is
thus used to model a maintenance or repair activity which replaces or effectively
overhauls the component or the dominant contributing component piecepart. Relatively
simple equations for the component unreliability and unavailability result when the good
as new model is used.
The Goodas OldRestoraionModel
The good as old restoration model is the other straightforward model which can be used
to model a testing, inspection, maintenance, or repair activity. The good as old
restoration model is the model opposite to the good as new restoration model with regard
to renewal. The good as old model provides the following answers to the three questions
regarding 1) the component state 2) the age effect, and 3) the aging effect:
1. The component is in an up state after the activity,
35
2. The age remains the same after the activity (no renewal) and there is no age
setback,
and
3. There is no change in the aging behavior after the activity; the same aging
failure rate applies with the same age after the activity as before the activity.
The good as old restoration model, or good as old model for short, is used to model an
activity which assures that the component is in an operational state (up state) but does
minimal repairs on the component if the component is not functionally failed. Thus,
there is no removal of degradations or no major renewal of the component and the age of
the component basically remains the same. When applied to a component piecepart, then
the piecepart is not renewed or significantly refurbished. The good as old model is used
to model a surveillance test or inspection activity which does not involve major
maintenance or repair actions. If a failure or severe degraded state is detected which
requires correction then the repair activity can be separately modeled as a good as new
activity using the good as new model. Thus, routine surveillance testing and inspection
can be modeled as being as good as old activities and when failure or severe degradation
occurs then the repair or corrective activity can be modeled as being a good as new
activity. The good as old model again produces relatively simple equations for the
component unreliability and unavailability.
To utilize the formulas for the good as new restoration model or the good as old
restoration model one must determine whether the aging control of a given test,
maintenance, or replacement is best described as being good as new or as being good as
old. The good as old model can be used to provide an upper bound and the good as new
model a lower bound for the effect of the activity. Combinations of the good as old and
good as new models can be used for different activities carried out at a plant. For
example, as was indicated, a surveillance test carried out to assure a component is
operational can be modeled as having minimal aging control unless the component is
found failed. The surveillance test can thus be modeled as being as good as old with the
repair of a failure being modeled as good as new. The surveillance test interval can be
used as the good as old restoration interval and the good as new restoration interval can
be taken to be the mean time to failure. If scheduled replacements or overhauls are in
addition carried out then these activities can separately be modeled. Thus, the good as
new and good as old models can provide flexibility in modeling the aging control of
testing and maintenance activities.
36
2.7 More Complex Maintenance and Restoration Models
The good as new and good as old models are useful for providing a first order evaluation
of the aging control of a test or maintenance activity. They can also provide bounds on
the aging control effects. More complex maintenance and restoration models involve
modeling the testing or maintenance activity to not be good as old or as good as new, but
somewhere between. More complex models can also involve modeling inefficiencies
associated with a test or maintenance. These more complex models can be applied when
detailed test and maintenance data is available. These more complex models can also be
useful for sensitivity studies to determine the sensitivity to the details of the test or
maintenance activity. For example, a useful analysis is to use the good as old and good
as new models to identify the risk important test and maintenances and then to carry out
more detailed sensitivity analyses on the risk important tests and maintenances. The
more complex models which will be covered are:
1) Modeling the effect of tests or maintenances on individual pieceparts of the

component,
2) Modeling the effect of tests or maintenances which only cover specific failure
causes or mechanisms,
3) Modeling explicitly the age setback which results from a maintenance or

repair activity,
4) Modeling the aging control of preventative maintenances in reducing the

aging rate,
and
5) Modeling inefficiencies associated with a test or maintenance activity.
These more detailed models are discussed in the subsequent sections.
Modeling Tests or Maintenanceson IndividualPieceparts
When modeling tests or maintenances on individual pieceparts, the component is

subdivided into its pieceparts and each piecepart is treated as a separate component.
Each piecepart can then be evaluated using the good as old or good as new model or
37
more complex models. For example, a valve can be divided into the valve driver (the
actuator, motor etc.), the valve interior (valve disk, etc.), and the valve exterior body. A
surveillance test which only tests a given piecepart can be modeled as being good as old
and an overhaul or repair of the piecepart can be modeled as being good as new, the
models now being applied to the specific piecepart.
The actual modeling involves replacing the component failure rate by the sum of its
piecepart failure rate contributions (e.g., in fault tree terminology replacing the
component by an "or gate" of its piecepart contributions).' The failure rate for each
piecepart then needs to be determined and the appropriate test and maintenance model for
each piecepart needs to be determined. When the good as old model is used for a
surveillance test and the good as new model is used for repairs the information required
is the appropriate test interval on the piecepart and the mean time to failure for the
piecepart. The test and maintenance procedures and historical data are used to identify
the specific pieceparts which are tested or maintained and the test and maintenance
frequency and characteristics. To prevent the models from becoming overly large, only
the major components can be expanded into their pieceparts. For components which
have one piecepart as the major contributor to the failure mode, the component may be
treated as being equivalent to the piecepart.
Modeling Tests or Maintenanceson Specific FailureCauses
If a test or maintenance only covers specific failure causes or failure mechanisms then
the activity can be modeled as only controlling a portion of the component failure rate.
The portion or fraction of the failure rate which is controlled is determined based on the
fraction of total failure contributions which is covered by the test or maintenance. This
fractional restoration modeling is similar to the previous piecepart modeling except
instead of pieceparts being identified, the fraction of failure causes or mechanisms which
is controlled is modeled.
To apply the fractional restoration approach, the total component aging failure rate is
multiplied by the fraction of failure contributions which is controlled by a given test or
maintenance. This fractional failure rate contribution (the total failure rate times the
*The piecepart contributions can more accurately be modeled as being competing cause contributions,
however the difference from the summation of the contributions, which ignores the interactions is
generally insignificant
38
fraction) can then be assigned to be good as new after the given failure causes or
mechanisms are corrected. Surveillance tests which only detect this fraction of failure
contributions can be assigned to be as good as old for the fractional contribution. The
failure rate can be divided into different fractional contributions and each separately
tracked. Failure causes or mechanisms not detectable are treated as not being testable or
maintainable.
Explicitly Modeling Age Setback
A partial restoration of a component can be explicitly modeled as a partial age setback,

where the component age is setback to a given value. In modeling age setback, the age
reduction after a given repair or maintenance activity is explicitly determined. For the
good as old model, there is no age setback i.e., the age reduction is zero after the activity,
but the component is assured to be an up state. For the good as new model, there is a
complete age setback, i.e., the age reduction is equal to the age of the equipment before
the activity, resulting in as good as new equipment For a repair or maintenance activity
modeled as having an intermediate age setback, the value of age setback needs to be
estimated. For a given age setback of Aw after an activity the age of the component is
reduced from w to wI where
W = w -Aw (34)
where w is the component age before the activity. The aging component failure rate then
begins at wI after the activity for the calculation of the component unreliability and
component unavailability.
To calculate the component unreliability or component unavailability versus time, the

component age versus time needs to be tracked, accounting for age setbacks by repairs
and maintenances. The formulas in the appendix can be used to calculate the
unreliability and unavailability with appropriate setback of the age using Equation (34).
With regard to applications, the more difficult question is the appropriate age setback
value to assign to a given repair or maintenance activity. Little work has been done in
this area. Sensitivity studies can be performed by varying Aw to determine the
sensitivity to the degree of restoration. Data analyses can also be carried out to estimate
the age setbacks associated with different types of maintenances by constructing
appropriate likelihood functions with the age setback the parameter to be estimated.
39
Modeling the Effects of PreventativeMaintenancesin Reducing the Aging Rate
Preventative maintenances may not restore an aged part, but instead may slow the aging
process of the component. Preventative maintenances which involve lubrication or
cleaning are of this type and can be modeled as reducing the aging rate of the equipment.
Before the preventative maintenance the aging rate has a value say of a and after the
maintenance, the aging rate is reduced to a, where
a = a - Aa (35)
and where Aa is the aging rate reduction due to the preventative maintenance.
To determine the aging control of a preventative maintenance the aging rate reduction
needs to be related to the component failure rate change. If the linear aging failure rate
model is used (Equations (5)-(9)) then the aging rate reduction Aa can be directly
interpreted as the change in the linear aging rate a. The new linear aging rate after the
preventative maintenance is then al. Preventative maintenances may be more generally
carried out to control continually increasing rates of aging. To determine the aging
control on aging rates which increase with time, nonlinear failure rate models are
required.
For a nonlinear aging failure rate model, the aging rate a is more generally defined to be
the rate of increase in the failure rate with age:
a =d (36)
dw
where the right hand side of Equation (36) is the derivative of the failure rate with age.
The aging rate a is thus generally a function of the age w, i.e., a = a(w) and can increase
with age.
From the above relationships, a change in the aging rate can be related to the change in
the failure rate, which is what is required for application in an APSA. If Aa is modeled
for a preventative maintenance activity, then the new aging rate can be determined from
Equation (35), and Equation (36) can be used to determine the failure rate after the
preventative maintenance. With the failure rate after the activity determined, the
40
component unreliability and unavailability can then be calculated using the equations in
the appendix.
For example, for the Weibull failure rate model (Equations (10)-(14)) the failure rate
derivative is
dA 2b b-i
dw- wb (37)
Equating the derivative to the new aging rate a, after the preventative maintenance gives
the component age w, after the preventative maintenance,
Aob b-i (38)

wo
or
1w (39)
PA1b J39
The new age w, can then be used for the Weibull failure rate to calculate the unreliability
and unavailability of the component. The more difficult task is to estimate the aging rate
change Aot for a given preventative maintenance activity. Perhaps the best use of the
model is to carry out sensitivity analysis to determine the importance of preventative
maintenance actions. This preventative maintenance model can also be used to analyze
maintenance data to estimate the aging rate reductions for given preventative
maintenances using appropriate likelihood functions with Aa as the unknown parameter
to be estimated.
Modeling Test and MaintenanceInefficiencies
Finally, sometimes a test or maintenance may be inefficient in detecting or correcting a

degradation or failure. One of the most straightforward ways to model a test or
maintenance inefficiency is to increase the test interval or replacement interval to be an
"effective" interval. The use of an effective interval models the inefficiency as being
associated with a constant probability per activity of not detecting or correcting the
41
degradation or failure. Let eT be the test efficiency which is defined to be the probability
of detecting a failure at a test. Let T be the actual test interval and T. be the effective test
interval which is the expected test interval at which the failure is detected. Then using
standard probability relationships*,
T
Te = (40)
eT
Similarly if eL is the probability of a restoration activity effectively restoring the

component, L the actual restoration interval, and L8 the effective restoration interval,
then
Le =-. (41)
eL
The effective intervals can then be used in the formulas for the component unavailability
in place of the actual intervals. Sensitivity studies can be performed by varying er or eL
to determine the risk impact of inefficiencies. Those tests or maintenances which are
most sensitive are those where the efficiencies need to be the highest.
Maintenance and repair inefficiencies can also be modeled using the partial restoration
model or age setback model previously given. Instead of assuming good as new
restorations these partial restoration models can be used to model partial restorations
associated with the maintenance or repair. Sensitivity studies can be performed to
evaluate the impacts of partial restorations instead of complete restorations.
2.8 Summary of Models and Data Needed to Ouantifv Component Reliability and
Unavailability Effects of Aging
The previous sections described the failure models and test and maintenance models
which can be used to quantify the reliability and unavailability of a given aging
component. The data required for these models were also discussed. It is useful to
*For a detection probability of er per test the average number of tests before detection is 1/&r based on
the standard geometric probability distribution. For an interval of T between tests, the average interval to
detection is then T/CT.
42
- - -
summarize these modeling and data needs since this encompasses what is needed to
transform a PSA into an APSA.
In summary, for each component for which aging effects are to be explicitly quantified
an aging reliability model is needed. This consists of:
1. A model of the age-dependent failure rate for the component. Commonly used
age-dependent failure rate models are:,
The linear failure rate model,

The Weibull failure rate model,
or
The exponential failure rate model.
The formulas for the above failure rate models were given in Section 2.2. For a given
age-dependent failure rate model the parameters of the model must be estimated from
failure rate data or from engineering information. Instead of the age dependent failure
rate, a model of the overall time trend in the failure rate or in the failure frequency can be
used to show overall trends due to aging. However, an overall time trend model cannot
be used with different test and maintenance models since the overall time trend has the
test and maintenance effects already incorporated and the aging failure rate is not
separated out as a function of age.
In addition to the age-dependent failure rate model when the aging control effects of test
and maintenance practices are also to be explicitly evaluated then appropriate test and
maintenance models are required. Specifically,
2. A model of the age control of each test, maintenance, or repair performed on the
component or component piecepart is required. The most straightforward models
are
The good as old restoration model,

and
The good as new restoration model.
43
More complex models can also be used, including piecepart models, fractional
restoration models, age setback models, and preventative maintenance models.
Inefficiencies in an activity can also be modeled. For a given test or maintenance model,
the parameters in the model need to be determined from the procedures and from
historical data if available. For the good as old or good as new model the only required
input data is the interval at which the activity is performed. Sections 2.6 and 2.7
described the different test and maintenance models. The appendix gives formulas for
calculating the component reliability characteristics using the aging failure rate model
and the appropriate test and maintenance models. These reliability characteristics are
then used to transform the PSA to an APSA. The next chapter, Chapter 3, describes the
approaches which can be used for this transformation process.
44
3. APPROACHES FOR TRANSFORMING A PSA INTO AN AGE-
DEPENDENT PSA
3.0 Introduction
There are different approaches that can be used for transforming a PSA to an age-
dependent PSA, or an APSA. We describe three basic approaches. The first approach
involves carrying out a standard PSA evaluation a number of times with different
component failure rates used in each evaluation. The different component failure rates
which are used are stepwise approximations to time dependent component failure rates.
This approach is straightforward, but the age of each component cannot be separately
tracked. Hence, the effect of testing, maintenance, or repair on the age of the component
or on the aging rate cannot be explicitly modeled.
The second approach for transforming a PSA into an APSA is to substitute age
dependent component models into the PSA quantification formulas. The fundamental
PSA formulas for the core damage frequency in terms of the component failure
probabilities and component unavailabilities are still used. However, the quantification
formulas for the component failure probabilities and componentyunavailabilities are
changed from the usual, steady state formulas to those which explicitly incorporate
aging. This approach efficiently calculates the core damage frequency and system
unavailabilities as a function of plant age. However, the aging contributors are not
resolved in detail and die calculations can be time consuming if sensitivity studies are
performed.
The third approach for transforming a PSA to an APSA is to calculate appropriate risk
importance coefficients from the standard PSA and to combine these with calculated
component aging effects using separate component aging models. This approach is
effective for resolving in detail aging contributors, including all the aging interactions.
The approach also allows sensitivity studies to be efficiently carried out. However, the
approach can be time-consuming calculationwise. Formal uncertainty analyses can also
be time consuming.
The following sections describe each of the above approaches. The features of each
approach are further described. Each approach is described in sufficient detail to allow
implementation of the approach for specific applications.
45
3.1 Successive Stepwise Evaluations Using a Standard PSA
The approach of successively quantifying a standard PSA to approximate aging effects is

straightforward. The time dependent failure rate of a component is approximated by a
series of stepwise constant failure rates as shown below.
Approximation of a Time Dependent Failure Rate by a Stepwise
Constant Faihre Rate
F
a
u
r
R
t
to t2 Time t t4
For a time interval with a given approximate constant failure rate, the component can
thus be treated as in a standard PSA which assumes the component has a constant failure
rate. For multiple components which have time dependent failure rates, the time
dependent failure rate of each component is approximated by a series of constant failure
rates using the same set of time intervals for each component. A standard PSA
evaluation (computer run) is then made for each time interval using the associated
constant failure rate for each component. The results from the quantifications for
different intervals then serve as a stepwise approximation of the time dependent core
damage frequency as illustrated below.
Core Damage Frequency at Different Tis Points
Core Damage Frequenc
tI t 2Tm t3 t4
46
The advantage of the successive stepwise approach is that standard PSA models and
standard.PSA computer codes can be used. Care must be taken in selecting the width of
the time intervals, though this is a numerical accuracy problem which can be addressed
by using finer time intervals.
The basic disadvantage of using the stepwise, approximate approach is that individual
component ages cannot be explicitly tracked since a common time must be used for all
components. Thus, the effects of test, maintenance, or repair on the age of the
component and on the aging rate cannot be explicitly modeled. When a component is
replaced, repaired or maintained, the failure rate value after the activity is the same as the
failure rate value before the activity. This is because of the constant failure rate
assumption used in the PSA run.
The implications of using the same failure rate before and after an activity depend upon
the type of component failure rate function which is used as the basis for the histogram
approximations. When overall time trend failure rates (or failure frequencies) are used as
the basis for the histogram approximations, then the effects of testing, maintenances, and
repair are already factored into the failure rates. The effects of testing, maintenance, and
repair thus cannot be separated out and be explicitly modeled. Different testing,
maintenance, and repair programs consequently cannot be evaluated because of the built-
in effects of the testing, maintenance, and repair program under which the time,
dependent failure rates were collected. The PSA results then provide a best-estimate,
stepwise approximation of the time dependent core damage frequency and system
unavailabilities under the given testing, maintenance, and repair programs built into the
failure rates.
If age dependent component failure rates are instead used as the basis for the histograms,
where test, maintenance, and repair effects have been removed, then the PSA runs
generally will provide a conservative, upper bound evaluation of the aging impacts. All
the component ages are equated to a common running time in using the common set of
time intervals for the histogram approximations. Thus, it is assumed that components are
never overhauled or replaced with new components. Because of the constant failure rate
assumption, surveillance tests, maintenances, repairs, and replacements are thus all
effectively modeled as being good as old. No component is ever replaced with a new
component and no degradation is ever removed from a component, causing a renewal of
the component. Because of the good as old assumption for all activities, the stepwise
47
approximation approach in this case generally provides a conservative, upper bound
evaluation of the age-dependent core damage frequency and system unavailability.
Thus, it is important to know whether the basic component failure rate functions which
are used are time trends and incorporate the effects of testing, maintenance, and repair or
whether they are age dependent failure rates which separate out the effects of testing,
maintenance, and repair. This was the basic aging failure rate issue that was discussed in
Section 2.3 and is important for the proper understanding and application, not only for
the approximate stepwise PSA approach, but for the other approaches for carrying out an
APSA as well. It is an issue since as was indicated in Section 2.3, failure rate data bases
may not clearly differentiate as to whether the failure rates are overall trends or are age
dependent with the test and maintenance effects removed. Compounding the problem,
data analyses may calculate overall time dependent failure rates and erroneously call
them age dependent.
One final point involves the PSA minimal cut sets. * A truncated list of minimal cut sets
which have been evaluated as being the most important is among the standard results
provided by the PSA. A list of minimal cut sets is used not only in the successive
stepwise approach, but in all approaches to quantify the aging effects. Truncation of the
minimal cut sets by the PSA can lead to underestimating of the aging effects because
minimal cut sets which were unimportant in the original PSA can become important
when aging is considered. This can particularly be the case for minimal cut sets which
contain components which are simultaneously aging. The simultaneous aging will cause
a multiplicative increase in the minimal cut set contribution to the core damage
fiequency. To check on the effect of additional minimal cut sets being considered, the
minimal cut sets can be expanded to include the multiple components which are aging
and the evaluations carried out again using the expanded set. If this checking is not done
then it should be made clear as to what contributions are included and which are not.
This applies to all the approaches. A summary of the basic features of the stepwise
approach is given in Table 5.
*Aminimal cut set is a standard PSA term and isa smallest combination of component failures with an
initiating event that will cause a core damage event
48
TABLE 5. THE SUCCESSIVE STEPWISE APPROACH
FOR PSA AGING EVALUATIONS
Basic Procedure: Approximate the time-dependent component failure rate by a series of

constant stepwise failure rates in given time intervals. Use the same time intervals for all
components. For a given time interval select the appropriate constant failure rate for
each component and quantify the PSA as usual. Repeat for each time interval of interest.
Advantages: Standard PSA models and software can be used with no modifications
required.
Disadvantages: Because of the constant failure rate assumption in the PSA, the age of
the component and the aging rate are never modified. Thus, different test, maintenance,
and repair programs which affect the aging or aging rate cannot be evaluated.
Special Points: If overall, time dependent (time trend) failure rates or failure
frequencies are used for the histogram fittings then the effects of tests, maintenances, and
repairs are built into the failure rates. The PSA results then give a best estimate, stepwise
approximation to the core damage frequency for the aging data and given test,
maintenance, and repair programs built into the failure rates. If age-dependent failure
rates are used with the test, maintenance, and repair effects removed then all tests,
maintenances, and repairs are effectively treated as good as old because of the constant
failure rate assumption. This will generally give conservative, upper bound core damage
frequency results within the histogram approximations. Truncation of the minimal cut
sets, can lead to underestimation of the aging effects if omitted minimal cut sets become
significant because of aging effects.
49
3.2 Substitution of Agingf Models Into a PSA
A second approach for transforming a PSA into an APSA is to substitute aging models
into the PSA component quantification formulas. Instead of calculating component
unavailabilities and component unreliabilities using standard PSA constant failure rate
models, appropriate component aging reliability equations are substituted to calculate
age-dependent component unavailabilities and unreliabilities. The same fault tree and
event tree models in the PSA are
used. However, different values are calculated for the component unavailabilities and
unreliabilities using appropriate component quantification formulas as given in the
appendix. The component unavailabilities and unreliabilities need to be calculated for
each of the times of interest to obtain the time dependent core damage frequency. If
average aging effects are of interest, then averages are calculated over appropriate time
intervals.
To understand the substitution ofthe component aging models into the PSA, consider the
standard PSA formula for the core damage frequency C in terms of the component
contributors:
N (2
C= Ii q qi2.qimi (42)
i=1
where
Ii- = the initiating event frequency for a given (43)
accident
and
qil qi2.qim, = the product of component unavailabilides *
in the ith minimal cut set (a combination of (44)
component failures causing a core damage
event if the initiating event occurs)
* A component unreliability is used instead of a component unavailability when the failure mode is
failure to rn, instead of failure to start.
50
The minimal cut sets can vary in size, and to be general there are assumed to be mj
components in the ith minimal cut set. Also, to be general it is assumed that N minimal
cut sets are obtainable from the original PSA. The unavailabilities can also include
human error contributions. The same general formula as Equation (42) also applies for
system unavailabilities, expected health consequences, and other results calculated by the
PSA.
Equation (42) is dependent only upon the basic logic structure of the PSA and doesn't
depend on any specific quantitative models used for the initiating event frequencies and
component unavailabilities. In standard PSA evaluations, constant failure rate models
are used to calculate values for Ii and c4 2*- qjn.. For age-dependent evaluations,
age-dependent models are instead used to calculate these values at different time points
or plant ages. The calculations need to be repeated for each time point of interest,
however the same basic PSA formula, Equation (42), is used with only different numbers
substituted forIi andqi q(*-- qii..
In substituting age-dependent models for the component unavailabilities and

unreliabilities (or initiating event frequencies) the following poihts need to be considered
for implementations:
1. Truncation of minimal cut sets can again cause certain aging contributors to be
omitted. A truncated minimal cut set contribution can become particularly
important if multiple components in the minimal cut set are aging, multiplying
the contribution of the minimal cut set. The truncation effect can be investigated
by expanding the minimal cut sets to include those omitted minimal cut sets
containing aging components.
2. If different test and maintenance policies are to be evaluated, then age-dependent

component failure rates need to be used which separate out the effects of tests,
maintenances, and repairs. Appropriate test, maintenance, and repair models of
age control, such as those given in Chapter 2, should be used to calculate the
component unavailabilities.
51
3. If time dependent (time trend) failure rates or failure frequencies with built in
test, maintenance, and repair effects are used then different test, maintenance, and
repair practices cannot be evaluated. Good as old restoration models should then
be used in the component quantifications to maintain the time dependent
behaviors in the failure rates.
The advantage of the substitution technique is its straightforwardness. Standard PSA

models and minimal cut sets are used with the quantification formulas modified for the
component unavailabilities and unreliabilities (or initiating event frequencies). Separate
computer subroutines or modules can be developed to calculate the age dependent
component unavailabilities and unreliabilities which can then be used in the standard
minimal cut set expressions for the core damage frequency and system unavailabilities.
Sensitivity and uncertainty analyses can also be incorporated into the quantifications.
The disadvantage of the substitution technique is its limited resolution of aging

contributors. The final core damage frequency and system unavailabilities are
determined but the aging contributors are not delineated. The minimal cut set
contributions give the total contributions from component combinations, but the aging
effects are not separated out. How much a component's aging contributes to the core
damage frequency and how much component aging interactions contribute are not
identified. Also, the calculations at different time points can be tedious if many
sensitivity calculations are carried out, although efficient calculational algorithms can be
constructed to help address this problem. Table 6 summarizes the features of the
approach for substituting aging models into a PSA.
3.3 The Risk Importance Approach for Evaluating Aging Effects
The third approach for transforming a PSA into an APSA is the risk importance approach
whose methodology is described in NUREG/CR-5510(7). Appropriate risk importances
are calculated from the standard PSA and are then combined with component aging
models to give the core damage frequency increase due to aging. Other risk importances
can also be calculated from the PSA to give the increase in any risk result due to aging,
such as increases in system unavailability due to aging. Even though the approach is
described in NUREG/CR-5510, it is summarized here in context of its use for
transforming a PSA into an APSA.
52
TABLE 6. THE SUBSTITUTION APPROACH FOR PSA AGING
EVALUATIONS
Basic Procedure: Use aging models to calculate the component unavailabilities (or
unreliabilities) at a given time point (or given plant age) for the aging components.
Alternatively, use the aging models to calculate average increases in unavailability due to
aging. Substitute the component unavailability values into the PSA minimal cut set
equations to obtain the age-dependent core damage frequency. Repeat at different time
points for time dependent evaluations.
Advantages: The standard PSA logic models and minimal cut sets can be used with
only the quantification formulas modified for the component unavailabilities,
unreliabilities and initiating event frequencies. Separate subroutines can be set up to
calculate the age-dependent unavailabilities, unreliabilities, and initiating event
frequencies to replace the standard PSA quantifications. Different test and maintenance
programs can be evaluated for their risk effectiveness if appropriate age dependent
failure rates are used.
Disadvantages: The aging effects are not delineated and hence prioritization of the
aging effects can not be readily carried out. The calculations can be tedious and time
consuming if different aging calculations are performed such as for sensitivity
evaluations
Special Points: Age-dependent component failure rates should be used which separate
out the test and maintenance effects if different test and maintenance programs are to be
evaluated. If time dependent failure rates or failure frequencies are used which
incorporate test and maintenance effects then good as old restoration models should be
used to maintain the time dependent failure rates or failure frequencies. Truncation of
the minimal cut sets can lead to underestimation of the aging effects and can be expanded
to determine the impacts. The contributors included and those onitted should be
described.
53
Let AC be the increase in the core damage frequency due to aging, i.e. the difference
between the core damage frequency with aging and without aging. Then AC can be
expressed as a sum of contributions,
AC= SjAq + S AqjqjAqj (45)

i j>i
+ YSjk Aqi qj qk ...

k>j>i
+I S12..n AqlAq2 ... Aqn

n
where
Aqi = the increase in the unavailability of (46)

component i due to aging (the difference
between the unavailability with and without
aging)
and where Si, SiQ,... S12...a are appropriate risk importance coefficients determined from
the standard PSA. NUREG/CR-5510 describes how to calculate the risk importance
coefficients and provides algorithms which can be used for implementations. The last
term on the right hand side is the contribution from the largest size minimal cut sets
obtainable from the PSA (without truncation). The total core damage frequency C with
aging can be obtained by adding AC to the core damage frequency without aging
calculated in the PSA C., i.e., C = C. + AC Similar expressions can be written for the
aging increase in any other risk result.
Equation (45) is an exact expression and gives a detailed breakdown of all the aging
contributions to the core damage frequency increase due to aging. Each Aqi is the
difference between the unavailability q calculated with aging and the unavailability q.
calculated without aging in the PSA, Aq = q - q 0. The unavailability increase Ach can be
time dependent (or age dependent) or can be an average aging effect over given time
periods or replacement intervals.
54
With regard to the detailed breakdown of contributions, the first term on the right hand
side of Equation (45) gives the sum of the contributions from individual component
aging effects:
Si Aci = the core damage frequency contribution (47)

from the aging of component .
The second term gives the sum of the two component interaction contributions from two
components simultaneously aging:
Sij A4 Aqj - the core damage frequency contribution (48)

from the simultaneous aging of components
i andj.
The higher order interaction contributions for three components simultaneously aging,
etc. are given by the succeeding terms on the right hand side of Equation (45) up to the
maximum interaction contribution, which is the size of the largest minimal cut set
modeled in the PSA.
Besides breaking down the core damage frequency impact due to aging into the detailed
individual and interaction aging contributions, each contribution in Equation (45) shows
the risk importance factor and the aging effect. For an individual component
contribution Si A, Si is the standard PSA core damage frequency importance of the
component and Ac4 is the aging effect. Similarly, for an interaction contribution, the first
factor is the core damage frequency importance of the interaction and the second factor,
the product of Aq%'s, is the aging effect. For example, for the two component interaction
contribution SjAqAqj, Sq is the risk importance of the contribution and AqiAqj is the
aging interaction effect. Thus, one can determine how much of each core damage
frequency contribution is due to the core damage frequency importance of the
components and how much is due to the aging of the components. The core damage
frequency importances of the components are determined from the basic PSA and
reflects the basic design and operation of the plant. The aging effects Aq, depend upon
the component aging failure rates and the aging control of the test and maintenance
programs.
55
In applications, the contributions to AC are often truncated to only consider second or
third order interactions, i.e. the terms on the right hand side of Equation (45) are
truncated at the second or third summation. As indicated, NUREG/CR-5510 gives
algorithms for calculating the risk importance coefficients SiSij, etc. The aging effects
Aq are calculated using appropriate aging reliability models as has been previously
described. If the aging control of test and maintenance programs are to be explicitly
evaluated for their risk effectiveness, then again age-dependent component failure rates
which separate out the effects of testing, maintenance, and repair should be used.
Appropriate formulas for Aq%are given in the appendix.
The disadvantage of the above aging-risk importance approach is that the risk importance
coefficients Si, Sij, etc. can be tedious to calculate, particularly if many terms are
determined for the contributions to AC Efficient algorithms for determining the risk
importance coefficients focused on those components which are aging can help address
this problem. Uncertainty and sensitivity evaluations can also be tedious if uncertainties
and sensitivities in the risk importance coefficients are included. Equations and
algorithms for the risk importance coefficients given in NUREG/CR-5510 can be used to
obtain the uncertainties and sensitivities in terms of basic component data uncertainties.
The advantage of the aging risk importance approach is that it provides a detailed
breakdown of all the aging contributions to the core damage frequency increase due to
aging. How much of the contribution is due to the risk importance of the contributor and
how much is due to the aging effect is also given. Pioritization of the aging contributors
and of test and maintenance activities thus can be effectively carried out using the
approach. From an implementation standpoint, because the aging evaluations Aqi are
separated from the PSA risk importances S., Si etc., the PSA needs to be solved only
once to determine the risk importances, regardless of the number of aging evaluations
carried out using different aging rates, test and maintenance models, etc. Because the
focus of this report is on aging prioritizations and sensitivity studies, the risk importance
aging approach will be principally used in the applications to be discussed. Table 7
summarizes the features of the risk importance aging approach.
56
TABLE 7. THE RISK IMPORTANCE APPROACH FOR PSA AGING
EVALUATIONS
Basic Procedure: Use the standard PSA to calculate the risk importance coefficients (S.,
Sq, etc.) for each component and combination of components which are aging. Multiply
the coefficients by the increases in unavailability (Aq, Aqj, etc.) due to aging and add to
obtain the total core damage frequency change (AC). Recalculate the unavailability
increases for the time points or ages of interest for time dependent evaluations.
Advantages: The detailed aging contributions are identified and prioritized. The risk
importance factor and the aging effect factor for each contribution is identified. The risk
importance coefficients need only be calculated once for different aging evaluations
which are carried out.
Disadvantages: Calculation of the risk importance coefficients can be time consuming

if many higher order coefficients are calculated. Including uncertainties and sensitivities
in the risk importance coefficients can be tedious.
Special Points: Age-dependent component failure rates which separate out test,
maintenance, and repair effects should again be used if the aging control of test and
maintenance programs are to be explicitly evaluated. If the summation of terms for AC
is truncated at some term then the effects of the truncation can be examined by
calculating additional terms. The truncation limit should be documented as part of the
results.
57
4. APPLICATIONS OF AN AGE-DEPENDENT PSA
4.0 Itrmoduction
An age-dependent PSA, which we have called an APSA, can be used for various
applications. A particular application of an APSA can be classified according to the
principal focus of the analysis, which can be:
Evaluation of the core damage fiequency with aging effects incorporated,
Evaluation of test and maintenance effectiveness in controlling aging impacts,
Prioritization of aging contributors to the core damage frequency,
or
Evaluation of the sensitivities and uncertainties in risks due to aging effects.
When the focus is on evaluation of the core damage frequency with aging effects
incorporated then the emphasis is on determining the bottom line number for the core
damage frequency and system unavailabilities with aging effects explicitly modeled.
When the focus is on evaluation of test and maintenance effectiveness in controlling
aging effects then the emphasis is on evaluating the effectiveness of test and maintenance
programs in controlling the risk impacts of aging under specified aging behaviors.
Different test and maintenance strategies may be investigated and different plausible
aging failure rates may be used. When the focus is on prioritization of aging
contributions then the evaluations are geared toward the ranking and screening of aging
contributors. Finally, when sensitivity and uncertainty analyses are the principal focus
then variations and distributions in the risk are determined for variations and distributions
in aging behaviors.
The next section identifies the basic questions that need to be answered in setting up an
APSA calculation for any of the above applications. These questions and their answers
provide a framework for carrying out one or more of the above applications. These
questions address the aging data, the test and maintenance models, the PSA information
and the results from the APSA.
58
4.1 Ouestions To Be Addressed in Setting Up an APSA Application
There are four basic questions that need to be answered in setting up an APSA for any
specific evaluation. These questions are:
1. What is the available component aging failure rate data?
2. What is the available test and maintenance information?
3. What is the available type of PSA information?
and
4. What are the aging results of most interest?
The answers to the questions will determine the type of APSA application which can be
best carried out and the type of APSA approach which is most suitable. Figure 6
presents these questions in terms of a decision tree with the possible answers. The
following sections discuss in more detail the alternative answersto each of these
questions and the implications of these answers.
4.2 Impact of Available Component Aging Failure Rate Data
The component aging failure rates which are available can be of three types:
Plant specific aging failure rates,
Generic aging failure rates,
or
Subjective estimates of aging failure rates.
59
FIGURE 6. QUESTIONS TO BE ANSWERED IN SETTING UP AN
APSA EVALUATION
Available aging failure rate data?

I
I I
plant
specific generic subjective data
I
Available test and maintenance information?
I
detailed
I I
minimal
information information
I
Available PSA information?
r ~~~~~~~~~~~~~~~~~~~~~I
I
event trees and
fault trees
I
I
truncated
minimal
cut sets
Results of most interest?
I
I
bottom line pndoitization
numbers and sensitivities
60
Plant specific aging failure rates are the most precise data and are based on failure
histories of components at the plant being analyzed. The component failure rate model
(e.g. linear, Weibull, or exponential) can be selected which is most consistent with the
data, and the parameters of the model can be estimated using statistical data analysis
approaches. Plant specific aging failure rates are, however, generally not available, or
are available for only selected components.
Generic aging failure rates are aging failure rates which have been obtained from a
population of plants. The generic aging failure rates represent average aging failure rates
over the components which are combined in the data base. In performing uncertainty
analyses or carrying out sensitivity studies, it is therefore important to include possible
plant specific variations. Generic aging failure rates are potentially obtainable from
existing generic data bases such as the Nuclear Plant Reliability Data System (NPRDS),
if times of failures are recorded. Incomplete records and possible errors in failure
classification in these data bases need to be considered as part of any uncertainty or
sensitivity analyses which are performed.
The last type of data consists of subjective estimates of aging component failure rates.
Because of the lack of plant specific aging data and even generic aging data, subjective
estimates of aging failure rates will often be the data which is most available. An
example of this type of data is the data base which is utilized in NUREG/CR-5510 (7).
Subjective estimates of aging failure rates involve translating engineering information
and experience about a component into an equivalent aging failure rate estimate. If
subjective aging failure rate estimates are used then it is important that sensitivity or
uncertainty analyses be also carried out to evaluate uncertainties and variabilities
associated with the subjective data.
The type of aging data which is principally available is an important factor in

determining the most meaningful focus of application for an APSA. The type of aging
data which is principally available also is an important factor in determining the
meaningful interpretations which should be placed on the results from the APSA. If
plant specific data is principally available then all the different focuses of applications of
an APSA discussed in the previous section are equally meaningful because accurate
aging data relevant to the plant is available. The APSA results furthenore can be
interpreted as being directly applicable to the plant within the associated data
uncertainties. All the different approaches for transforming a PSA to an APSA can be
61
effective with the substitution approach most efficiently providing the bottom line core
damage frequency value and system unavailability values. The risk importance aging
approach is most effective if detailed aging contributors are of prime interest, broken
down into the individual component contributors and the contributions from aging
interactions. Finally, the successive approximation approach can provide quick bounds
or approximations to the aging impacts.
If generic data is principally available then all applications can be carried out but the
results must be interpreted as under the limitation of generic data being used. The most
meaningful applications are those which account for possible plant specific variations by
varying the aging rate data to account for plant specific effects. Results which are not
sensitive to these aging rate variations are the most meaningful. Prioritizations of aging
contributors and sensitivity studies, along with uncertainty analyses, are thus among the
more meaningful applications. With regard to APSA approaches, the successive
stepwise approach is effective for efficiently bounding or approximating the aging
effects. The risk importance aging approach is effective if aging contributor
prioritizations are the focus. The substitution approach is useful if the focus is on final
core damage frequency values. For all these approaches, the effects of plant specific data
variations about the generic data values must again be included.
If subjective aging failure rate estimates are principally available then the aging failure
rate data can have especially large uncertainties. For subjective aging failure data, the
most meaningful applications involve prioritizations of aging contributors and sensitivity
analyses of aging impacts. The meaningful results are those which are account for the
large uncertainties in the subjective aging failure rate data. Extracting meaningful results
thus includes focusing on the order of magnitude of the value and focusing on
relationships between values. The relationships which can be focused on include
relationships between sizes of core damage frequency impacts and sizes of aging rates,
relationships between types of maintenance and resulting aging impacts, and
relationships between types of aging contributors and their aging rates. The risk
importance approach can be useful for identifying these relationships as well as for
prioritizing aging contributors in terms of order of magnitude impacts. The other APSA
approaches can also be useful for these types of results.
62
4.3 Impacts of Available Test and Maintenance Information
There are two levels of test and maintenance information which can be available:
Minimal information, basically providing only the type of activity performed and
its frequency,
or
Detailed historical information providing the times the activity was performed,
the type of activity, degradations monitored and corrected, and pieceparts
replaced.
When minimal information is only available then one basically only knows the type of
activity, i.e. whether a surveillance test, maintenance, or replacement was performed, and
the average frequency of the activity. Preventative maintenances may be differentiated
from corrective maintenances. The frequency of the activity may also only be known
approximately. When only minimal information is available then the basic, good as new
and good as old restoration models described in Chapter 2 are the most meaningful
models which are consistent with the minimal test and maintenance information. These
models are first order models and can be used to obtain a first order evaluation of the
aging impacts.
As described in Chapter 2, the good as new model treats the test or maintenance activity
as basically removing all significant aging effects and restoring the component to as good
as new. The good as old model treats the test or maintenance activity as assuring the
component is operational but as carrying out no major renovations of the component so
as not to affect the age of the component or its failure rate. As was indicated,
surveillance tests which principally check the operational status of the component can be
modeled to be as good as old restoration activities. Maintenances which correct
degradations and repairs can be modeled to be as good as new restoration activities. The
good as old and good as new restoration models require only the frequency of the activity
and hence data requirements are minimal. Sensitivity studies can be carried out by
changing various good as new models to good as old and vice versa. Also, the partial
restoration models previously described in Chapter 2 can be used for sensitivity studies to
study the effects of partial restorations between good as old and good as new.
63
When detailed historical information on tests and maintenances is available then such
information as the time of each test or maintenance, the specific actions that were carried
out, the degradations and component conditions that were observed, and the pieceparts
that were replaced are obtainable. This means that, first of all, information is available to
determine the applicability of the good as old or good as new model for a particular
activity. Information is also available to determine whether preventative maintenances
need to be separated from corrective maintenances as discussed in Chapter 2.
Information may also be available to determine the applicability of the partial restoration
models or piecepart maintenance models described in Chapter 2.
For the more complex test and maintenance models, the parameters of the model can be
determined from the information which is available on the activity. The parameters may
be estimated from engineering information and data analyses. The maintenance data can
be analyzed to not only estimate the parameters, but also to determine the most
applicable model.
When more detailed information is available on the test and maintenance activities, any
of the approaches for transforming a PSA to an APSA can be effective depending upon
the focus of the evaluation. The successive stepwise approach will be useful when only a
bound or first approximation for the resulting core damage frequency is desired. The
substitution approach will be effective when the traditional PSA results are desired but
now with aging effects incorporated. The risk importance approach is most useful for
prioritization and sensitivity studies.
4.4 Impact of PSA Information Which is Available
The type of PSA information which is available affects the completeness and detail to
which aging contributors can be evaluated. The PSA information which is available can
be:
The complete PSA event tree and fault tree models
or can be:
A restricted PSA set of minimal cut sets.
64
For each of these cases, it is assumed that the PSA has been developed to a basic
component level i.e. to an individual valve, pump, and relay level. If the PSA has only
been developed to a grouped component level or to a train level in which individual
components afe not identified then component aging effects cannot be evaluated by
transforming the PSA. For each of the above cases of information, it is also assumed that
the data used for the PSA quantification is also available, which includes the initiating
event frequencies, component failure rate data, the test and maintenance data, and the
human error data.
All the approaches for transforming a PSA to an APSA generally use as basic input the
PSA minimal cut sets which are generated from the event trees and fault trees. In the
first case above, when the complete event tree and fault tree models are available then the
minimal cut sets can potentially be generated to the level needed to include all the aging
component contributors. In particular, the minimal cut sets can be generated to include
all significant aging contributors including minimal cut sets containing multiple aging
contributors.
In the second case, when only a restricted PSA set of minimal cut sets is available then
only those components in the restricted minimal cut sets can be evaluated for their aging
contributions. As was indicated previously, a set of minimal cut sets is usually generated
by the PSA as the contributors which are most important. The set of minimal cut sets is
truncated to contain those minimal cut sets whose probability is above some cutoff value
such as lx108. If this truncated set of minimal cut sets is only available then certain
aging contributions may be neglected, especially multiple aging components in the same
minimal cut set which can significantly increase the cut set probability when aging is
considered.
Consequently, when a restricted set of minimal cut sets is only available then particular
individual component aging contributions and particular multiple component aging
contributions (aging interactions) may be neglected which can change the results. For
any application it is thus important to specifically define the contributions which are
included for the aging evaluations. This applies to the case where any truncated set of
minimal cut sets is used, even when the event trees and fault trees are used to generate an
expanded, but still truncated set of minimal cut sets. The criteria used for selecting the
minimal cut sets and the contributors included and excluded should be carefully
documented.
65
Finally, in addition to the level of PSA information, the scope of contributors included in
the PSA is important since it determines the scope of aging contributors which are
included in the APSA. A standard PSA generally does not consider individual balance of
plant (BOP) contributors, lumping all pertinent contributors into a transient initiating
frequency value. Thus, the effects of aging of BOP contributors cannot be evaluated
using this PSA. PSAs also do not generally include many passive component
contributors such as piping and cables. The PSA should be expanded to include these
other contributors, when impacts from these contributors are desired. In all cases, the
scope of the PSA should be documented.
4.5 Impact of Results Which Are of Most Interest
Finally, for a particular application, there can be specific results which are of most
interest, with other results being of lesser importance. All the APSA approaches can
provide a complete set of risk results. However, as was previously noted, a given
approach is most efficient in providing certain types of results. The successive stepwise
approach is best at providing relatively quick approximations or bounds on the aging
impacts. The substitution approach is best at providing bottom line values for the core
damage frequency and system unavailabilities. The risk importance approach is best at
providing prioritizations of aging contributors and sensitivity results.
As was also previously indicated, the desire for specific results must be tempered with
the availability of aging failure rate data, test and maintenance data, and PSA
information. When minimal data exists then the selected models and the APSA approach
used need to account for the lack of data. Bottom line core damage frequencies and
system unavailabilities can be obtained but uncertainty analyses and sensitivity analyses
need to accompany the results. With little data, the most meaningful results are
sensitivity analyses, prioritizations of aging contributors, and relationships determined
which are not sensitive to specific data values. This applies for any APSA approach
used.
66
5. PRIORITIZATIONS OF AGING CONTRIBUTORS
5.0 Intrduction
This chapter illustrates the applications of an APSA for the objective of prioritizing aging
contributors. The prioritization of the contributions to the core damage frequency from
aging active components will be specifically evaluated. The same methodology could be
applied in prioritizing passive component contributions, provided the PSA contains the
passive components and aging failure rates can be obtained for the passive components.
For the prioritization of active components demonstrated here, a PSA will be used which
has already been developed for a plant. The particular PSA which is used is not of
concern since the focus here is on the basic approaches for using an APSA to prioritize
aging contributors.
For the prioritization of aging contributors, the approach which will be used for
transforming a PSA into an APSA will be the risk importance approach for evaluating
aging effects previously described in Section 3.3. As was described in Section 3.3, the
risk importance approach provides a detailed accounting of the aging contributors
including all the aging interactions. The substitution approach for evaluation aging
effects described in Section 3.2 could also be used, however the prioritization of the
contributors would be in terms of the minimal cut sets instead of the detailed individual
and interaction contributors. The successive stepwise approach could also be used to
produce first approximations of the contributors but again they would be in terms of the
minimal cut sets which contribute to the core damage frequency.
The following sections describe specific steps that need to be carried out in applying an
APSA for prioritization evaluations. Considerations involved in selecting aging failure
rate models and aging failure rate data are described. Selection of test and maintenance
models is also discussed. Finally, organization and grouping of the results for
prioritization applications is discussed. Even though the specific focus is on using the
risk importance approach for prioritizing aging contributors, the considerations apply to
any APSA approach used for prioritizing contributors.
-67
5.1. Application of the Risk Importance Aging Approach
As was discussed in Section 3.3, in the risk importance aging approach, the core damage
frequency increase AC due to aging is expressed as a sum of contribution terms from
successively higher order aging interactions;
AC = AC + AC 2 +...+ACn (49)
Where AC1 is the contribution from individual component aging effects, AC2 the
contribution from two component aging interaction effects, etc. The maximum size n of
the interactions is the largest core damage minimal cut sets obtainable from the PSA.
For prioritization applications, we must first decide upon the maximum size of
interactions to consider. Considering all interactions will be exact but can be time
consuming in calculating all the contributions. The higher order interaction terms can
also be repetitive in that the same components will appear in the contributions.
Considering only single component aging effects AC1 will include the dominant
individual components which individually contribute most to the core damage frequency
effects from aging. We shall consider interactions up to second order to include in the
prioritization those additional components which individually may not be important but
which jointly can be important aging contributors. Even though we consider only second
order interactions, the approaches and guidelines we describe are applicable for any size
interactions considered.
When interactions up to second order are considered then Equation (49) for AC
approximates to
AC = AC + AC2 (50)
The first order contribution AC1 is the sum of the individual component aging
contributors;
AC1 = Acj (51)

i
68
where Aci is the core damage frequency contribution from the aging of component i. As
indicated in Section 3.3, Aci can be expressed as a product of the risk importance Si of
the component and the aging effect aging am% on the unavailability of the component,
Aci =Siqi (52)
Similarly, the second order contribution AC2 is the sum of the joint component aging
contributors,
AC 2 = yaC1 (53)
j>i
where Acij is the interaction contribution from the simultaneous aging of components i
and j. As was indicated in Section 3.3. the contribution Ic&can be expressed as a
product of the joint risk importance Sij of the component combination and the effect of
simultaneous aging Ach Aqj on the component unavailabilities,
Acij = SijAqiAqj * (54)
All the individual aging contributors and second order contributors are thus identified in
this joint prioritization evaluation. Additional, higher order terms AC3, etc. can be
generated to check whether any new contributors appear.
A personal computer program was constructed to calculate the risk importance

coefficients Si and S4 from the core damage frequency minimal cut sets supplied by the
PSA using the algorithm given in the NUREG/CR-5510(7). All the data in the original
PSA (failure rates, test intervals, initiating event frequencies, etc.) are used to calculate
values for Si and Sij. The aging effects Aq, are calculated with reference to the baseline
values determined in the PSA. The core damage frequency effects AC1 and AC2 are thus
calculated with reference to the baseline core damage frequency in the PSA. To obtain
the total core'damage frequency including aging, one can add the aging contribution AC
to the baseline PSA value. As was indicated in Section 3.3, the advantage of the risk
importance approach is that the original PSA may be used to calculate the risk
importance coefficients, which need to be done only once for as many different aging
analyses that may be carried out which only involve changing Aq4 or Aq%.
69
The core damage frequency increase AC (including AC, and AC 2 ) which will be
calculated will be the average increase due to aging between replacements (renewals) of
the component. The average core damage frequency increase is calculated by using the
average unavailability increase Aq between component replacements in Equations (52)
and (54). If the component is not replaced during the plant lifetime then the average
increase over the plant lifetime is used. The average increase applies to any year. The
average increase AC added to the baseline core damage frequency from the PSA will
give the new, average core damage frequency at any year. The appendix gives the time
dependent unavailability formulas which can be averaged to obtain the average Aq.
5.2 Selection of Aging Failure Rate Models and Data
When plant specific failure data is available then the appropriate failure rate aging
models and data would be selected based on this plant data. To determine the aging
models and data for these models, the failure data would need to include at minimum for
each component of interest, the times of failure and times of major overhauls or
replacements. Using statistical approaches, the appropriate failure rate model, e.g.
whether linear, Weibull, or exponential, would be determined by fitting the alternative
models using likelihood or Bayesian approaches.
In lieu of having available plant specific data, generic aging failure rates or subjective
estimates of aging failure rates need to be used. Generic or subjective estimates of aging
failure rates can be useful in providing an initial prioritization of the aging contributors to
help initially focus maintenance activities, monitoring activities, and data collection
activities. When generic or subjective aging failure rates are used then sensitivity studies
or uncertainty analyses can carried to investigate effects of data variations and
uncertainties. he results from the sensitivity or uncertainty analyses can then be
factored into the prioritization conclusions. For the application here, the subjective,
aging failure rate data base in NUREG/CR-5248 (8) is used. This aging failure rate data
base is often called the TIRGALEX data base referencing the committee which had
oversight on the work.
The TIRGALEX aging rate data base assumes a linear aging failure rate model,
70
(55)
where (w) is the age dependent failure rate as a function of the component age w, X. is
the underlying constant failure rate as used in the PSA and a is the component aging rate
(the symbol "a" is used here instead of "t" in chapter 2). For the TIRGALEX data base,
a panel of experts estimated the aging rates a for various classes of components. Though
no systematic illicitation or estimation techniques were used, there was an attempt to be
consistent and to effectively utilize available engineering knowledge.
The TIRGALEX aging rates a are given in Table 8 for the active components in the PSA.
The TIRGALEX aging rates are intended to be subjective, generic aging rates.
NUREG/CR-5248 warns about the uncertainties associated with the values. As a means
of assessing the uncertainties in the TIRGALEX aging rates, NUREG/CR-5510 (7)
compared the TIRGALEX aging rates with aging rates which were obtained from
samples of Licensee Event Report (LER) data, data from the Nuclear Plant Reliability
Data System (NPRDS), and plant specific data. For diesels, pumps, and motor operated
valves, the median aging rate values from the samples of data agreed within
approximately a factor of 10 with the associated TIRGALEX values. For check valves
the TIRGALEX aging rate value was a factor of 10 to 100 lower than specific plant
estimates, principally because of check valve backleakage problems at specific plants.
Thus, the TIRGALEX aging rates generally agreed to within an order of magnitude with
the aging rates estimated from plant data and were biased low when plants had specific
problem components.
The TIRGALEX aging rates were generated in NUREG/CR-5248 for prioritizing

research needs. Thus, the TIRGALEX aging rates are best suited for prioritization
analysis as is carried out here. Because the aging rates are based on subjective judgments
and have large uncertainties, sensitivity analyses are also carried out for the prioritization
applications here. For the sensitivity analyses, two different aging rate data bases are
constructed for the components which were dominant PSA contributors. One sensitivity
aging rate data base represents low aging rates and one represents high aging rates.
Further details of the steps involved in generating the aging rates are provided in
NUREG/CR-5510.
Table 9 gives the low sensitivity aging rates for those components identified as being
dominant contributors in the baseline PSA. The aging rates are low aging rates in that
71
TABLE 8. TIRGALEX AGING RATES USED FOR ACTIVE COMPONENTS
COMPONENT AGING RATE (per hour per year)

Ac Bus 1.OE-09
Air-Operated Valve 4.OE-07
Battery 3.OE-07
Check Valve 4.OE-09
Circuit Breaker 2.OE-08
DC Bus 1.OE-09
Diesel Generator 3.6E-06
Motor-Driven Pump 2.OE-07
Motor-Operated Valve 3.6E-06
Relay 3.OE-07
Safety/Relief Valve 7.OE-07
Transformer 2.OE-09
Turbine Driven Pump 3.OE-06
Solenoid-Operated Valve 6.7E-07
72
the aging rate only doubles the baseline PSA failure rate of the component after 40 years.
All aging rates for other components not shown in the table are kept at their TIRGALEX
values. Table 10 gives the high sensitivity aging rates that are defined for these same
components. All other component aging rates are kept at their TIRGALEX values.
These high sensitivity aging rates are the same as the upper threshold aging rates used in
NUREG/CR-5510 and cause an unavailability of 0.1 after 18 months, which is a
significantly high unavailability and which is characteristic of severe aging.
One final comment should be made regarding the use of the linear failure rate aging
model given by Equation (55). Use of the linear aging model, instead of the Weibull
model or exponential model, ignores nonlinear aging effects exhibited by the
components. The linear aging model can thus be viewed as a first order linear
approximation for the aging behavior of the component. Prioritizations using linear
aging rates can consequently be viewed as first order prioritizations of the risk
importances of component aging effects. Investigations of the impacts of nonlinear
aging effects could then be carried out using appropriate nonlinear models.
5.3 Selection of Test and Models
If test and maintenance effects are to be explicitly incorporated when aging contributors
are prioritized, then appropriate test and maintenance models must be selected. For the
prioritization we shall assume technical specifications are followed at the plant with no
additional scheduled maintenance being carried out. This can be termed the minimal
maintenance situation. The prioritization of contributors which are obtained can then be
used to help focus and structure a risk-based aging maintenance program. Prioritization
of aging contributors under an existing scheduled maintenance program would require
selection of the applicable maintenance models. The prioritization results would then
identify what contributors would need further emphasis and what contributors could be
relaxed in terms of their maintenance activities.
For the prioritization, we thus assume surveillance tests are good as old. That is, if
degradations are detected at a test they are not removed as long as the component is still
functional. We also assume that when a component is found to be functionally failed at a
surveillance test that it is replaced or is repaired so that it is as good as new. This can be
an optimistic assumption if minimal repairs ae only carried out. The assumption
73
TABLE 9. LOW SENSITIVITY AGING RATES USED FOR DOMINANT
CONTRIBUTORS
Battery 2.OE-08
Check Valve 4.OE-09
Diesel Generator 2.0E-07
Motor-Operated Valve 2.OE-07
TABLE 10. HIGH SENSITIVTY AGING RATES USED FOR DOMINANT

CONTRIBUTORS
Battery 1.OE-05
Check Valve 1.OE-05
Diesel Generator 1.OE-05
Motor-Operated Valve 1.OE-05
74
can be chocked by treating repairs to also be good as old.(See the next chapter.) We
assume the same test intervals as used in the basic PSA. We do not assume that testing is
carried out more frequently if aging occurs. The applicable test interval for each
component will be shown in subsequent result printouts. In addition to the surveillance
test interval, we also need to assume efficiencies for the surveillance tests. From Section
2.7 again, the efficiency of a surveillance test is the probability that a given failure is
detected by the test.
Two sets of efficiencies are used for the prioritization evaluations. For one set of
evaluations, the surveillance tests are assumed to have 100% efficiency (efficiency
fraction T = 1). For the second set of evaluations, the surveillance tests are assigned the
efficiencies &rgiven in Table 2.6 of the TIRGALEX report NUREG/CR-5248. The
efficiencies are reproduced in Table 11. The APSA results from both these efficiency
data sets will be factored into the prioritizations.
5.4 Selection of the Formulas for the Component Aging Impacts
As additional input information required for the prioritization application, we need

formulas for Aq, the aging effect on the component. More specifically, Aq is the increase
in component unavailability over the base PSA value without aging effects;
Aq = q-qO (56)
where q is the component unavailability explicitly including aging and q. is the value
calculated in the base PSA assuming no aging (i.e. assuming a constant component
failure rate).
The formula for Aq, or equivalently for q, for each aging component is required. These
formulas are required, not only for the risk importance approach we are using here, but
also for the substitution approach. For the successive stepwise approach, age-dependent
formulas are not needed, but time steps would need to be selected which apply to all
components and which are used to approximate the time dependent or age dependent
component failure rates for each component
75
TABLE 11. TIRGALEX SURVEILLANCE TEST EFFICIENCIES
COMPONENT TEST EFFICIENCY

Ac Bus 0.45
Air-Operated Valve 0.45
Battery 0.86
Check Valve 0.09
Circuit Breaker 0.45
DC Bus 0.45
Diesel Generator 0.27
Motor-Driven Pump 0.44
Motor-Operated Valve 0.60
Relay 0.18
Safety/Relief Valve 0.82
Transformer 0.62
Turbine Driven Pump 0.44
Solenoid-Operated Valve 0.82
76
If we were also including aging effects on initiating event frequencies (e.g. on the pipe
rupture frequency) then we would also need formulas for the change in initiating event
frequency AX due to aging. However, we will not consider aging effects on initiating
event frequencies since our focus is on prioritizing active safety components. It is useful
to note however that the aging effect AX is simply
=Al o.0 -; (57)
where Xis the age dependent failure rate (i.e. initiating event frequency) such as the
Weibull failure rate and A, is the constant value used in the PSA. For the core damage
frequency contributions from aging, AX would then simply replace Aq in Equation (45).
We will prioritize the aging contributors with regard to their average aging effects on the
core damage frequency. The average aging effect for a component is the average
increase Aq due to aging between replacements or renewals of the component.* From
NUREGICR-5510 (7), for a linear aging failure rate, the average unavailability increase
Aq due to aging is
Aq= a(L-T)T aT2 (58)

4 6
where a is the aging rate, L is the replacement interval (or overhaul interval), and T is the
surveillance test interval for the component. The above formula models a surveillance
test as being good as old. The above formula can also be obtained by substituting a
linear aging failure rate into the general formula given in the appendix.
If components are replaced at failure, as we assume, then the replacement interval L for
each component is the mean time to failure for the component plus the time until the
failure is detected. The time until the failure is detected is, on the average, one half the
interval between tests. Since the test interval is generally small compared to the mean
time to failure we shall ignore the additional time until detection.
*If we were to prioritize the contributors for a given time, i.e. at a given plant age, then &qwould be the
time dependent unavailability increase. We would need to now the times at which the tests or
maintenances are performed in addition to the intervals.
77
Using the PSA constant failure rate value, the mean time to failure is simply one over the
component failure rate. The value for the mean time to failure for each component is
given as part of the results presented in the subsequent section. Calculation of the mean
time to failure using the PSA failure rate ignores the effects that the additional aging rate
has on the mean time to failure. For the baseline and low sensitivity aging rates, because
the aging rates are relatively small compared to the constant failure rates, these effects
are small, particularly when compared to the uncertainties in the failure rates. For the
high aging rates, neglect of these effects will cause the mean time to failure to be larger
than if these effects were considered. This will add an extra conservatism to these upper
bound evaluations.
Consequenty , we assume
L= (59)
Ao
where 34 is the PSA constant failure rate.
If there are no surveillance tests expected on the component between replacements, then
T is set equal to L in Equation (58) for Aq, as discussed in the appendix. As previously
discussed in Section 2.7, inefficiencies in surveillance tests can be modeled by
interpreting T in Equation (58) to be the effective test interval:
T= Q (60)
ST
where eT is the test efficiency.
Finally, we need to consider the case where the mean time to failure of the component is
larger than the plant lifetime. We will assume an extended plant lifetime of 50 years.
Then from NUREG/CR-5510 or using the formulas in the appendix,
Aq = la[toT+Y3] (61)
78
where to is equal to 50 years. Whether to is set equal to 40, 50, or 60 years should have
little effect on the prioritization results. If there is no surveillance test expected in 50
years then the formula for the unavailability increase with no testing is,
jq I 2 (62)
where again t is 50 years. We thus have all the formulas required for the analysis.
5.5 Detailed Prioritizations of and Component Contributors
Tables 12A and B, 13A and B, and 14A and B present the prioritizations of component
contributors for the three cases which were analyzed, which are termed the base case, the
upper bound sensitivity case, and the lower bound sensitivity case. The base case utilizes
the TIRGALEX aging rates and the TIRGALEX surveillance test efficiencies. The
upper bound case utilizes the high sensitivity aging rates and the TIRGALEX test
efficiencies. The lower bound case utilizes the low sensitivity aging rates and assumes
surveillance test efficiencies of unity.
For each case analyzed, two tables are presented. The first table presents the core
damage frequency increase AC, for individual component aging effects. This gives the
prioritization of individual component aging effects. The second table presents the core
damage frequency increases AC2 for double component aging interactions. This gives
the prioritization of two component aging interactions. For example, Table 12A presents
the ranked individual component contributions for the base case and Table 12B presents
the ranked two component contributions. The total core damage frequency increase from
aging is the sum of the contributions from the two tables. The top 25 contributors are
given in each table, representing approximately 99% of the total contribution to AC1 and
to AC 2. The results are straightforwardly obtained using the risk importance approach,
previously described and are organized in a tabular format to explicitly detail information
on the aging contributors.
The first column in a given table is the rank of the contributor according to the impact of
the aging effect on the core damage frequency contribution. The core damage frequency
79
TABLE 12A. CORE DAMAGE FREQUENCY INCREASE FROM AGING ACTIVE COMPONENTS
PLANT A: BASE CASE, SINGLE CONTRIBUTIONS
Plant A. Single Cotributions

TRGALEX Aging Raues
TIRGALEX Testing Effciencies
TotalAC: 1.8E-04 /year
Test
Rank Component Name Seasitivity Aging Rate MTBF Aql Inval AC
Coefficient o/yr) (months) (monhs) (/year)
1 LPR-MOV-Fr-1862A 1.5E-04 3.6E-06 167 2.62E01 30 3.9E.05
2 LPR-MOV-FT-1860A 1.5E04 3.6E-06 167 2.6E-01 30 3.9E-05
3 LPR-MOV-Fr-1890A 1AE-04 3.6E-06 167 2.6E01 30 3.5-05
4 HPI-MOV-Fr-1350 6.72.05 3.6E206 167 2.6E-01 30 1.7E-05
S LPR-MOV-Fr-1862B 2.1E-05 3.6E-06 167 2.6E-01 30 5AE.06
6 OEP-DGN-FS-DG01 3.8E-04 3.2-06 72 1AE-02 4 53E-06
7 LPR-MOV-Fr-1860B 2.0E- 3.6E-06 167 2.6E-01 30 5.3E-06
8 0EP-DGN-FR-6HDGI 3.4E-04 3.62-06 72 1AE-02 4 4.82E06
9 OEP-DGN-FS-DG03 2.0E-04 3.6E06 72 1.4E-02 4 2.8E206
00 10 OEP-DGN-FS-DG02 2.02-04 3.6-06 72 1.4E-02 4 2.8E06
11 0EP-DGN-FR-6HDG3 1.9E-04 3.6E-06 72 AE4-02 4 2.7E06
12 OEP-DGN-FR-6HDG2 1.7E-04 3.6E-06 72 1AE4-02 4 2.5E06
13 PPS-MOV-FT-1535 9.5E-06 3.6E-06 167 2.6E-01 30 2.4E-06
14 HPIJCKV-Fr-CV225 2.1E-03 4.02-9 0 4.8-04 11 1.7E06
15 H-PICKV-Fr-CV25 2.1E-03 4.0-09 0 4.8E-04 11 1.7E.06
16 HPI-CKV-Fr-CV410 2.1E-03 4.0E49 0 4.8E-04 11 1.7E206
17 HPI-MOV-F-1I1SC 5.7E-06 3.6E-06 167 2.6E201 30 1.5E-06
18 HPI-MOV-Fr-1115D 5.7E-06 3.6E2-6 167 2.6E-01 30 1SE-06
19 HpI-MOV-FF-1115B 5.7-06 3.6E-06 167 242.01 30 1.SE-06
20 HPI-MOV-F-1115H 5.7E-06 3.6E-06 167 2.6E-01 30 1.5E06
21 LPR-MOV-FT-1890B 4.5E-06 3.6E206 167 2.6E-01 30 1.2E.06
22 PPS-MOV-FT-1536 3.4E-06 3.62-06 167 2.6E-01 30 8.8E-07
23 HPI-MOV-FT-1867D 2.9E-06 3.6E246 167 2.6E-01 30 7.5E-07
24 OEP-DGN-FR-DG01 5.0E-05 3.6E-06 72 1.E-02 4 7.2E-07
25 SIS-ACI-FA-SISA 1.8E05 3.0E-07 0 1.8E02 6 5.4E-07
TABLE 12B. CORE DAMAGE FREQUENCY INCREASE FROM AGING ACTIVE COMPONENTS
PLANT A: BASE CASE, DOUBLE CONTRIBUTIONS
PlM A: Double Contribution

TIRGAL Aging Rates
iRGALE Testing Efficiencies
Totd,&C 7.6E404 /e
Tast Test
Rank Component Name Sensitivity Aging Rat MP &q1 Inti Componen Name Aging Rate MBF Aq2 Inerval aC
_ ____________ Codficent W/yr() (months) (mond) hlyr) (mths) (months) 0fycar)
I HPI-MOV-Fr-1115B 19E403 3.606 167 2.6801 30 HPI-MOV-Fr-1115D 3.6-06 167 2.68.01 30 1.38.04
2 HPI-MOV-Fr-1IISC 1-98.03 3.6-06 167 216E;01 30 HPI-MOV-Fr-1115H 3.6E-06 167 2.6E01 30 1.3E04
3 LPR-MOV-Pr-189A 15f03 3.6-06 167 2.68.01 30 LR-MOV-FI-1890B 3.6E06 167 2.6E-01 30 1.0E04
4 LPR-MOV-P1860A 15E-03 3.6E6 167 2.8i01 30 LR-MOV-Ft-1860B 3.66-06 167 2.6E.01 30 1DE004
5 LPR-MOV-FT-1862A 1.5E03 3.6E-06 167 21;E40 30 LPR.MOV-FF.1860B 3.6E06 167 2.6E.01 30 1.0E04
6 LPR-MOV-Fr-1860A 103 3.6E46 167 2A.01 30 LPR-MOV-Fr-1862B 3.6E06 167 2.6E01 30 1.0E04
7 LPR-MOV-Pr-1862A 15803 3.6.06 167 2.6E.01 30 LPR-MOV-PT-1862B 3.6E-06 167 2.6E-01 30 1.0E04
8 SIS-ACr-FA-SISB 6.5E43 3.0E-07 0 1sm2 6 SIS-ACr-PASISA 3.0E47 0 1.8E02 6 5.9E-06
9 RT-ACr-FA4MS 1503 30.07 0 1.8802 6 RMr-ACT-FA-MSB 3.0£-07 0 1.8E02 6 1AE.06
10 OEPON-Ft-6HD03 5S.03 3.806 72 IAE-02 4 OEP-WN-FR G1 3.6E06 72 IAE02 4 .IE06
00
P"a 11 OEP-DGN-FS-DOOl 4.9E45 3.6E-06 72 IAE-02 4 OEP-DGN-FS-D02 3.6E06 72 IAE02 4 9.9E07
12 OEP-DWN-FS-DGO1 4.9E-03 3.6E46 72 IA.E02 4 OEP-DON-FS-D03 3.6806 72 IAE-02 4 9.9E-07
13 OEP-DWN-P-D01 4.0E03 3.6E-06 72 1AE42 4 OEP-DGN-RD02 3.6E-06 72 IAE02 4 8.1E-07
14 OEP-DINFS-DWOI 4.0E-03 3.68E06 72 IA 2 4 OEP-DGN-FR-6HDO3 3.6-06 72 IAE02 4 8.1E07
15 O£P-DON-FS-DGO2 408.03 36E.06 72 1A.E02 4 0P-DGN-FR-MHDOI 3.6E-06 72 1AE-02 4 8.1E07
16 OEP-DGN-FR-6HDO1 4.0E-03 3..06 72 IAE402 4 0EP-DN-FR-6HDG2 3.6E06 72 IAE42 4 8.1E07
17 OEP WN-PS-W03 3.9E43 3.6806 72 IAE02 4 OEP-DGN-FR6HDGI 3.rz.06 72 1.4E02 4 7.8E-07
18 OEP-DOWN-FS-DG01 5.E-03 3.6E06 72 1.4E02 4 MSS-SRV-00-SOSRV 7.0-07 22 3AE03 22 2AE-07
19 U.l-MDP-FS-SllB 1--03 2.0E07 86 5.98.04 2 IBMMOV-1862A
3.6-06 167 2.6E01 30 2.3E-07
20 LPI-MDP-FS-SllA 1E03 2.OE-07 86 5.8804 2 -MOV-Fr-1on 3.68.06 167 2.6801 30 23E07
21 LPI-MDP-FS-SllB 1.5E.03 2.0E-07 86 5.M804 2 LR-MOV--1860A 3.6E06 167 2.6E.01 30 2.3E.07
22 LPI-MDP-FS-SIIA 1.5E03 2.0E07 86 5.8844 2 LPR-MOV-PT-1862B 3.6E-06 167 2.6E01 30 23E07
23 OEP-DGN-FR-6HDGI 4.6803 3.6E06 72 IAE02 4 MSS-SRV-O SSRV 7.0E-07 22 3.4E03 22 2.2E07
24 PPS-MOV-FC-1536 2.9E-06 3.6E-06 167 2681 30 PPS.MOVC-1535 3.6E-06 167 2.6E.01 30 .9E07
25 OEP-DGN-PS-DW03 9.1E04 3.6E.06 72 1AE02 4 OEP-DGN-FR-DG01 3.68.06 72 1AE42 4 1.8E-07
PLANT A: UPPER BOUND CASE, SINGLE CONTRIBUTIONS
Pat A: Sing Contlbutions

High Aging Razcs
TIROALEX Tesdg Eficici es
Total AC_ .OE-02 /year
Test
Rank Component Namo Smwsiivity Aging Rate MIBF ql Interval AC
Coefficiet A/yr) (months) (months) /yer)
1 MPI-cKV-Fr.CV225 2.IE-03 lOE-05 0 7.OE-01 11 2.SE03
2 HPI-CKV-Fr-CV410 2.1E-03 1.OE-05 0 7.E-1 11 2.4E-03
3 HPI-CKV-FI-CV25 2.1E-03 1.E05 0 7.OE-01 11 2.4E-03
4 AOC-CKV-FI-CV145 5.OE-04 1.OE-5 0 7.0E41 11 5.8E-4
5 ACCCCV-FT-CV147 5.OE-04 .OE-05 0 7.OE-41 11 5.8E-04
6 ACC-CKV-FT-CV130 S.OE-04 1.OE-05 0 7.OE41 11 5.8E-04
7 ACC-CKV-FT-CV128 S.OE-04 1.OE05 0 7.OE01 11 5.8E-04
8 LPR-MOV-FT-1862A 1.5E-04 1.OE-5 167 5.lE401 30 7.8E-05
9 LPR-MOV-Fr-1860A 15E-04 1.OE05 167 S.lE41 30 7.8E45
10 LPR-MOV-FT-1890A 1.4E-04 1.OE45 167 5.lE-01 30 7.OE-05
00
11 HPI-MOV-Fr-1350 6.7E05 1.OE05 .167 S.1E01 30 3.4E-05
12 OEP-DGN-FS-DG01 3.8E-04 1.OE-05 72 3.9E02 4 1.5E-05
13 OEP-DGN-FR-6HDG1 3.4E-04 1.OE05 72 3.9E02 4 1.3E05
14 LPR-MOV-F-1862B 2.1E-5 1.OE-05 167 5.1E01 30 .1E-05
15 LPR-MOV-FT-1860B 2.OE-05 1.OE45 167 5.IE-41 30 1.OE-05
16 OEP-DGN-FS-DGO02 2.OE04 .OE-05 72 3.9E02 4 7.9E-06
17 OEP-DGN-FS-DG03 2.OE04 O-05 72 3.9E 02 4 7.9E.06
18 OEP-DGN-FR-6HDG3 1.9E04 1.E-05 72 3.9E-02 4 7.6E-06
19 OEP-DGN-FR-6HD02 1.7E-04 1.0E05 72 3.9E02 4 6.8E 06
20 PPS-MOV-FT-1535 9.5E06 LOE.05 167 .IE-01 30 4.8E-06
21 MSS-CKV-FT-SGDHR 4.1E06 1.OE-0S 0 7.OE-01 11 4.7E-06
22 HPI-MOV-F-ll15C 5.7E-06 1.OE-5 167 5.1E41 30 2.9E-06
23 HPI-MOV-Fr-Il 15D 5.7E-06 1.OE05 167 5.lE01 30 2.9E-06
24 HPI-MOV-F-1SE 5.7E06 .OE-05 167 5.lE01 30 2.9E-06
25 HPI-MOV-Fr-1115B 5.7E06 1.OE-05 167 5.1E-41 30 2.9E-06
PLANT A: UPPER BOUND CASE, DOUBLE CONTRIBUTIONS
Plan A: Double Contributions

High Aging Rats
TIRGALEX Testing Effiiencies
-__
_ _ ___ __ ___ __ __ __ ___ _ ___ __ __ ___ __ ___ _ ___ _ ___ _ ___ _ _ _ Total AC: 4.5E 03 1year
Test Test
Rank Component Name Sensitivity Aging Rate MlMF AqI Intetvl CompoentName Aging Rate MF Aq2 Interval AC
______ ~hrlyr)
Coefiet (months) (monds) _ _ (I r) (months) (months) year)
I HP-MOV-F-1115B ISE03 1.02-05 167 5.12.01 30 HPI-MOV-Frl.lSD 1.04-05 167 S.1E01 30 4.9E.04
2 -MOV-F-1115C .9E-03 1i02.5 167 5.E-01 30 HP-MOV-FT-I1IS1 1.0E-05 167 5.IE-01 30 4.9E04
3 HPK-CKV-0CV28 1.3-02 1.0E45 0 7.0E01 11 HPI-MDP-FR-1A24H 1.0EM 86 2.9E'02 2 4.4E04
4 LPR-MOV-FT-1860A 1.53 1.-05 167 5.1301 30 LPR-MOV-F-1862B 1.01105 167 5.IE-01 30 3.91.04
S LPf-MOV.PT-1890A E 4-03 1OiS 167 5.12-O1 30 -MOV-F-18B 10.0E5 167 5.1E-01 30 3.9E.04
6 LPR-MOV.IT-1860A 1.E-03 1.0E'0S 167 5.11.01 30 LPR-MOV-FT-1860B 1.0E5 167 .1E-01 30 3.91.04
7 LP-MOV-FT-1862A 15E-03 1.o0s 167 .M2-01 30 UR-MOV-Pr-1862B 1.05 167 5.1E-01 30 3.9E04
8 LPR-MOV-FT-1862A 1SE03 1MME40 167 S.1E-01 30 LPR-MOV-F-1860B 1M 167 5.IE01 30 3.9E.04
9 AFW.TDP-FS-FW2 5AE-03 1.0E.05 72 2A0 2 4 142 1i0205 0 7.0E01 11 1.54-0
10 HPi-CKV4O-CV258 4.0fi03 1.0E05 0 7.0fi41 It HnP-MDP-FR-IA6HR OE4S 86 2.92.02 2 1A.4
11 AFW-MDP-FS-W3A 3.-03 1.0E-05 86 29E02 2 AFW.CKV-OO-CV157 1025 0 7.0E-01 11 1.2E-04
12 AFW {X4-O.CV172 3.6-03 1.0.45 0 7M01 11 AFW-MDP-FS-FW3B 1.0E-0S 86 2.9E02 2 1.2E-04
13 PPS-MOV-FT-1535 1.1E04 1S.05 167 5.1"1 30 AFW.CV-C142 1.2-05 0 7.04-01 11 6.7E05
14 AFW-CKV-00.CV142 .lE204 1.005 0 7.021 11 PPS.MOV-FT-1536 1.05 167 5.1fi01 30 6.7E05
15 LPI-MDP-FS-S[IA 1.503 1.005 86 2.9E4-2 2 IPIKV4O-CVS8 1.0E5 0 7.0E01 11 5.1E05
16 LPI-MDP-FS-SIB 1-03 1.0E45 86 2.913-2 2 LPI-CV0-CVSO 1.02-05 0 7.0E-01 11 5.IE-05
17 MSS-SRVOSGSRV 7.92.04 10-M 22 4.91-02 22 AFW..CVOO6CV172 1.0E.5 0 7.0-01 11 4.52.05
18 AFW-CKV.00.CV157 6.4E05 1.04-05 0 7.0E41 4 11 PPS-MOV-PT-1536 1.0£M 167 5.1E01 30 3.84-05
19 AFW-CKV-00-CV172 6.4ES 1I05 0 7.02-01 11 P1S-MOV-1536 102M 167 5.1E-01 30 3.8E-05
20 PI-MOV-PT-1535 6.4E45 1.0E45 167 S.1E-01 30 AFW-CKV4XO-CVS7 1.0-O0 0 7.0E-01 11 3.8M05
21 PPS-MOV-PT-1535 6AE-05 1.02.05 167 5.1E01 30 AFWCIV-OO.CV172 1.02-0S 0 7.04-01 11 3.820
22 OEP-DGN-FS-D0I 6.-04 1.0-05 72 3.9E2 4 AFW.CKV4XO-CV172 1.02-05 0 7.0E-01 11 3.04-05
23 LPI-MDP-FS-SIIB 1.5-03 1.0£-05 86 2.92 2 LPR-MOV-FI-1860A 1.0M.5 167 5.1-01 30 2.2110
24 LPI.MDP-FS-SIIA 1.5503 1.04-05 86 2.913-02 2 j-MOV-FT-1862B 1.0S05 167 5.1-01 30 222.05
25 LPI.MDP-FS-SIIA 15203 1.0E-05 86 2.9E-02 2 LPR-MOVFr-1860B 1.0E5 167 5.1-01 30 2.2205
PLANT A: LOWER BOUND CASE, SINGLE CONTRIBUTIONS
Plant A. Single
Conuibutons
Low Aging Ratcs
Efficiency = 1
Total AC: 6AE-06 /year
Test
Rank Component Name Sensitivity Aging Rate MTBF hql Ineval AC
Coefficient (hr/yr) (months) (monahs) (/year)
i LPR-MOV-FT-1862A 1.5E04 2.OE-07 167 8.8E-03 18 13E-06
2 LPR-MOV-Fr-1860A 1.5E-04 2.0E-07 167 8.8E 03 18 1.3E-06
3 LPR-MOV-FI-1890A 1.4E-04 2.OE-07 167 8.8E-03 18 1.2E-06
4 HPI-MOV-Fr-1350 6.7E-05 2.OE-07 167 8.8E-03 18 5.9E-07
S HPI-CKV-Fr-CV225 2.1E-03 5.0E-09 0 5.SE-05 1 1.9E-07
6 HPI-CKV-Fr-CV410 2.1E-03 5.OE-09 0 5.5E-05 1 1.9E-07
7 HPI-CKV-FI-CV2S 2.1E-03 5.0E09 0 5.5E-05 1 1.9E-07
8 LPR-MOV-FT-1862B 2.1E-0S 2.0E-07 167 8.8E-03 18 1.8E-07
9 LPR-MOV-FI-1860B 2.0E-05 2.0E-07 167 8.8E-03 18 1.8E-07
10 AFW-TDP-FS-FW2 6.8E-05 2.0E-06 72 2.2E-03 1 1.5E-07
11 SIS-ACT-FA-SISA 1.8E-05 3.OE-07 0 3.3E-03 1 9.8E-08
12 SIS-ACr-FA-SISB 1.8E-05 3.0E-07 0 3.3E-03 I 9.8E-08
13 PPS-MOV-FT-1535 9.5E-06 2.0E-07 167 8.8E-03 18 8.4E-08
14 OEP-DN-FS-DG01 3.8E-04 2.0E-07 72 2E2-04 1 8.2E-08
15 OEP-DGN-FR-6HDGI 3.4E-04 2.0E-07 72 2.2E-04 1 7.4E-08
16 AFW-TDP-FR-2P6HR 2.6E-05 2.0E-06 72 2.2F-03 I 5.6E08
17 HPI-MOV-F-lllSB 5.7-06 2.0E-07 167 8.8E-03 18 SbE-08
18 HPI-MOV-FT-111SD 5.7E-06 2.0E-07 167 8.8E-03 18 S.OE-08
19 HPI-MOV-Fr-ll15C 5.7E-06 2.0E-07 167 8.8E-03 18 S.OE-08
20 HPI-MOV-FI-111SE 5.7E-06 2.0E-07 167 8.8E-03 18 5.0E-08
21 ACC-CKV-F-CV130 5.0E-04 5.OE-09 0 5.5E-05 1 4.6E-08
22 ACC-CKV-FF-CV145 5.0E-04 S.0E-09 0 5.5E-05 1 4.6E-08
23 ACC4CV-FT-CV128 5.0E-04 5.0E-09 0 5.5E-05 1 4.6E-08
24 ACC-CV-Fr-CV147 5.0E-04 5.0E-09 0 5.5E-0S I 4.6E-08
25 OEP-DGN-FS-DG02 2.0E04 2.OE-07 72 2.2E-04 1 4.3E-08
PLANT A: LOWER BOUND CASE, DOUBLE CONTRIBUTIONS
Plant A, Double Contributions

Low Aging Rates
Efrciency mI
Test Test
Rank Component Nane Sensitivity Aging Rat MIBP Aql Interal Component Name Aging Rate M F Aq2 InterMal C
coefdent Llr) (months) (monts) _ _ /yr) (monts) (monf) year)
SISB 6.5-03 3.0F507 0 3.3E.03 I SIS-ACr-PA-SISA 3.OE.07 0 33E3 1 1.9E507
2 HPI-MOV-F-1115C 19E03 2Afi07 167 8-03 18 HP-MOV-Fr-1115B 2.OE-07 167 81.03 18 1.S-07
3 HPI-MOV-FT-I 15B ISE-03 2.0547 167 88i503 18 HPI-MOV-FT-1115D 2.05-07 167 8.8E 03 18 15.07
4 LPR-MOV-Fr-1862A 1.5E03 2.E407 167 880 18 -MOV-PT-186OB 2.05-07 167 8.8E03 18 1.2507
S LPR-MOV-Fr-1862A 1.503 2.OE-07 167 8.8503 18 LPR-MOV-PT18B 2.OE07 167 88M03 18 1.7
6 LPR-MOV-Fr-1860A 1.5E-03 2.0E-07 167 8. 3 18 LPR-MOV-FT-186 2.OE407 167 8.8-03 18 125.07
7 LPR-MOV-FT-1860A 1SE53 2.0-07 167 8.8-03 18 LPR-MOV-PT-1860B 2.OE07 167 8.8E-03 18 1.2E.07
8 LPR-MOV-Fr.180A 1S0 2.05.07 167 8.503 i8 LP1-MOV -1890B 2.07 167 8.8E.03 18 12E07
9 RMT-ACT-AARM5 1.5E03 3.05-07 0 335.03 1 RMT-ACr-PAMS 3.0E-07 0 33E03 1 4.508
10 Lt-MDP-FS-SIlB 15E03 2.05.07 86 2.6E-0 I EPR-MOV-PT-180A 2.05.07 167 8.8503 18 3.4E09
11 Ul-MDP-FS-SIIA 1543 2507 86 2.65304 I LPlMOV-FT-1860B 2.0E-07 167 8.8E.03 18 3AE-09
12 ln-MDP-F5-SIIB 1E-03 2.0E07 86 2.E04 1 LPR-MOV-F-1862A 2.0E07 167 8.8E03 18 3.4E09
13 l-MDP-FS-SIIA 1.5E03 2.OE407 86 2.544 1 LR-MOV-F-18m 2.0E07 167 8E03 18 3.4E509
14 LPR-MOV.FT-162A 1.0E-03 2.0.7 167 8.M-03 18 LPIDP-FR-B21HR 2.007 86 2.6E04 1 2.3E9
15 LPR-MOV-F-1862B 1.0E03 2.OE.07 167 88.03 18 LPl-MDP-VR-A2IHR 2.0E07 86 2.6E-04 1 235.09
16 LPR-MOV-F1-1860B 1.0E03 2.05.07 167 8M-03 18 I-MDP.FR-A2IHR 2.007 86 2.6E.04 1 2.3E
17 LPR-MOV-FT1860A tLOE03 2.507 167 8M03.4 18 LPI-MDP-FR-B2IHR 2.5-07 86 2.6E-04 1 2.3E509
18 AFW-TDP-FR-2P6RR I.E04 2.0E06 72 225.03 1 A A-PA-M3B 3.047 0 335E03 1 2.1E-09
19 APW-TDP-FR-2P6HR 1.8E04 2.056 72 22 3 1 AW-ACr-PA-P3A 3.5-07 0 33-03 1 2.IE09
20 LPI-MDP-FS-SIIA 15.03 2507 86 265-4 1 SS-ACT-FA-SISB 3.0E07 0 33E-03 1 2.15E09
21 LPI-MDP-PS-SIIB 1503 2.0E07 86 2.04 1 SIS-ACrT-A-SISA 3.0E-07 0 33E03 1 2.1E09
22 UR-MOV-F1862B 5.0E-04 2.OE-07 167 88M3 18 2.0547
MMP-R- 86 2.6E-04
23 LPR-MOV-FT-1862A 5.0E04 2.0E-07 167 8.8E-03 1 l.E-09
18 -P-R-B 4 2.0547 86 2.6544 1 1.1E49
24 AFW-TDP-FS-FW2 SAE03 2.OE06 72 2-03 I AFW-CKV-00.CV142 5.0E-09 0 5.55.05 I l.E-09
25 AFW-TDP-FS-FW2 135i03 2.0E-06 72 22E03 1 MSS-SRv.MSGSRV 7.0EM 22 3.1E04 18 9.0E-10
contribution from every aging component and every combination of aging components in
the PSA is determined allowing the contributions to be ranked in detail. The second
column in the table is the name of the component as defined in the PSA. The component
name identifies the system, component type, failure mode, and specific component
identifier. The codes for the identifiers used in the PSA are given in Table 15.
The third column in a given table is the risk sensitivity coefficient, or risk importance
coefficient, Si or Sij for the contributor. The coefficient gives the core damage frequency
importance of the contributor as determined by the original PSA. Basically, Si or Sij is
the change in core damage frequency per unit change in the unavailability of the
contributor. The value of the risk sensitivity coefficient, or risk importance coefficient,
is determined by the basic design and operational procedures of the plant.
The next three columns give the component aging rate, component replacement interval,
and component test interval which determine the component unavailability increase Aq
caused by aging. The aging effect Aq is determined using the equations previously
given. Because we are assuming the component is replaced at failure, the replacement
time is the mean time to failure (denoted as MTBF in the tables)*. Where there is no
MTBF given, the mean time to failure is larger than 50 years and the formula for a
nonreplaceable component (Equation (61) or (62)) is used. The test interval given is the
effective test interval for the base case and upper bound case. For the lower bound case,
test efficiencies of 1 were assumed as was indicated.
It is useful to print out the aging rate, replacement interval, and test interval as shown to
identify the factors which determine the aging effect Aq. The core damage frequency
increase AC due to aging is the aging effect on the contributor multiplied by the risk
importance of the contributor. For the single component contributor, the aging effect is
simply Aq1 and AC is the product of the sensitivity coefficient Si and Aqc. For the double
component contributors, the aging effect is Aqi Aqj and AC is the product of the
sensitivity coefficient Sij and Aqi Aq.
The tables thus provide all the ranked contributors to the core damage frequency increase
AC due to aging. The risk importance factor (sensitivity coefficient) and the aging effect
factor which combine to give AC are explicitly given and the aging rate, replacement
* The mean time between failure (MBF) includes the component repair time, which isassumed to be
small compared to the mean time to failure (MlF. Hence MTFB-2 MITP.
86
TABLE 15. COMPONENT IDENTIFIERS
SYSTEM CODE
ACC - Accumulators
AFW - Auxliary Feedwater
HPI - High Pressure Injection
LPI - Low Pressure Injection
SIS - Safety Injection System
LPR - Low Pressure Recirculation
PPS - Primary Pressure Relief
OEP - Onsite Emergency Power
MSS - Main Steam System
COMPONENT CODE
ACT- Actuator
CKV - Check Valve
DGN - Diesel Generator
MDP - Motor Driven Pump
MOV - Motor Operated Valve
SRV - Safety Relief Valve
TDP - Turbine Driven Pump
FAILURE MODE CDE
FA - Fail to Actuate
FR - Fail to Run
FS - Fail to Start
FT - Fail to Transfer
00 - Fail to Open
87
interval, and test interval which determine the aging effect Aq are explicitly identified.
The tables can now be used to help focus aging control and aging management activities.
5.6 Groud Comnent Prioitizations
The detailed prioritizations in Tables 12, 13, and 14 can also be used to obtain grouped
prioritizations of the component contributors in which the components are grouped
according to various criteria. We will demonstrate the grouping approach by grouping
contributors which have the same order of magnitude contribution. Tables 16, 17 and 18
provide the grouped prioritizations of the contributors by order of magnitude of their
core damage frequency (AC) contribution. Te groupings of the contributors in the
tables thus focus on the general sizes of core damage frequency contributions. The
groupings also serve to account for uncertainties in the calculated contributions.
Table 16 groups the component contributors in Tables 12A and 12B for the base case by
order of magnitude contribution. Each component in a double contributor set in Table
12B is assigned to a group in Table 16 based on the size of the double component
contribution AC. Thus, the grouping combines the single and double contributors into
one overall prioritization. Tables 17 and 18 provide similar grouped prioritizations for
the upper bound case and lower bound case, respectively. From Tables 16, 17, and 18 it
is observed that the same contributors often appear even when different aging rate data
are used, although in some cases new contributors appear.
Table 19 combines Tables 16, 17, and 18 and gives the top order of magnitude
contributors from each of the three tables, then the second highest order of magnitude
contributors etc. Table 19 is thus a relative way of incorporating sensitivity results into
one prioritization. Table 20 is another way of combining the sensitivity study results into
one grouped prioritization by identifying all contributors above a given value of AC.
Table 20 is thus a threshold approach for combining the results from the sensitivity
studies. Other groupings of the contributors can also be carried out depending on the
specific objectives of the analysis.
88
TABLE 16. COMPONENTS PRIORITIZED BY ORDER OF
MAGNITUDE CONTRIBUTION: BASE CASE
Contributor Range of AC
1E-5 to 1E4
HPI-MOV-FT- 15B 0
HPI-MOV-FT-1 115C
HPI-MOV-FT-l1l SD
HPI-MOV-FT-1 115E
LPR-MOV-FT-1860A
LPR-MOV-FT-1860B
LPR-MOV-FT-1862A 0
LPR-MOV-FT-1862B 0
LPR-MOV-FT-1890A
LPR-MOV-FT-1890B
1E-6 to 1E-5
HPI-MOV-FT-1350 0
1E-7 to E-6
SIS-ACT-FA-SISA
SIS-ACT-FA-SISB 0
OEP-DGN-FS-DG0
OEP-DGN-FR-6HDGl 0
OEP-DGN-FS-DG02 a
OEP-DGN-FS-DG03
OEP-DGN-FR-6HDG3 0
OEP-DGN-FR-6HDG2
PPS-MOV-FT-1535
HPI-CKV-FT-CV225
HPI-CK V-FT-CV25 .
HPI-CKV-FT-CV410
RMT-ACT-FA-RMTSA
RMT-ACT-FA-RMTSB 0
89
TABLE 17. COMPONENTS PRIORITIZED BY ORDER OF MAGNITUDIE
CONTRIBUTION: UPPER BOUND CASE
- - - - - -
Contributor Range of AC Contributor Range of AC
IE-3 to 1E-2 IE-5 to IE-4
HPI-CKV-Fr-CV225 LPR-MOV-FT-1862A
HPI-CKV-FT-CV410 0
PPS-MOV-FT-1536
HPI-CKV-FT-CV25 PPS-MOV-FT-1535 0
LPI-MDP-FS-SI1B 0
IE-4 to 1E-3 LPI-MDP-FS-SI1A

ACC-CKV-F-CV147 LPI-CKV-OO-CV58
ACC-CKV-FT-CV145 LPI-CKV-OO-CV50
ACC-CKV-FT-CV130 MSS-SRV-OO-SGSRV
ACC-CKV-FT-CV128 HPI-MOV-FT-1350 0
HPI-MOV-FT-I 1 SE 0
OEP-DGN-FS-DG01
HPI-MV-FT- 11 5D CPC-MDP-FR-SWA24 0
HPI-MOV-FT-l 115C CPC-CKV-OO-CV113 0E6to1-
HPI-MOV-FT-1115B LPI-MDP-FR-B21HR
HPI-MDP-FR-1A24H 0 LPI-MDP-FR-A21HR 0
HPI-CKV-OO-CV258 0
OEP-DGN-FR-6HDG1 0
LPR-MOV-FT-1890B OEP-DGN-FS-D03
LPR-MOV-FT-1890A OEP-DGN-FS-DG02
LPR-MOV-FT-1862B OEP-DGN-FR-6HDG2
LPR-MOV-FT-1862A 0
LPR-MOV-FT-1860B PPS-MOV-FC- 1536

LPR-MOV-FT-1860A PPS-MOV-FC-1535
AFW-TDP-FS-FW2 OEP-DGN-FR-6HDG3 0
AFW-CKV-OO-CV142 LPI-MDP-FR-B24HR
HPI-MDP-FR-1A6HR LPI-MDP-FR-A24HR
AFW-MDP-FS-FW3B SIS-ACT-FA-SISB
AFW-MDP-FS-FW3A SIS-ACT-FA-SISA
AFW-CKV-O-CV172 MSS-CKV-F1-SGDHR
AFW-CKV-OO-CV157 OEP-DGN-FR-DG01
CVC-MDP-FR-2A1HR
HPI-MOV-FT-1867D
RMT-ACT-FA-RMTSB
RMT-ACT-FA-RMTSA
OEP-DGN-FR-DG03
OEP-DGN-FR-D&02
AFW-MDP-FS
90
TABLE 18. COMPONENTS PRIORITIZED BY ORDER OF
MAGNITUDE CONTRIBUTION: LOWER BOUND CASE
Range of AC
Contributor Range of AC
1E-7 to 1E-6
LPR-MOV-FT-1862A
LPR-MOV-FT-1860A
LPR-MOV-FT-1890A IE8to10
HPI-MOV-FT-1350 iE8toI-
SIS-ACT-FA-SISA
SIS-ACT-FA-SISB
HPI-CKV-FT-CV225 0
HPI-CKV-FT-CV25 0
IHPI-CKV-FT-CV410 0
LPR-MOV-FT-1862B
LPR-MOV-FT-1860B S
AFW-TDP-FS-FW2 0
HPI-MOV-FT-1 115B
HPI-MOV-FT- 1 15C
HPI-MOV-FT-1115D
HPI-MOV-FT-i l 1SE
LPR-MOV-FT-1890B
1E-8 to 1E-9
PPS-MOV-FT-1535 0
OEP-DGN-FS-DG01
OEP-DGN-FR-6HDG1
AFW-TDP-FR-2P6HR
ACC-CKV-FT-CV128
ACC-CKV-FT-CV130
ACC-CKV-FT-CV145
ACC-CKV-FT-CV147
RMT-ACT-FA-RMTSA 0
RMT-ACT-FA-RMTSB
OEP-DGN-FS-DG02
OEP-DGN-FS-D03
OEP-DGN-FR-6HDG3 4
OEP-DGN-FR-6HDG2 la
PPS-MOV-FT-1536
AFW-ACT-FA-PMP3A
AFW-ACT-FA-PMP3B
HPI-MOV-FT-1867D 0
91
TABLE 19. RELATIVE ORDERING OF THE COMPONENT
CONTRIBUTORS: COMBINATIONS OF THE THREE CASES
Relative Relative
Contributor Contribution Contributor Contribution
1st order of 3rd order of
HPI-CKV-FT-CV225 magnitude LPR-MOV-FT-1862A magnitude
HPI-CKV-FT-CV25 PPS-MOV-FT-1536
HPI-MOV-FT-I 15B LPI-MDP-FS-SIlB
HPI-MOV-FT-11 15C LPI-MDP-FS-SIlA
HPI-MOV-FT-Il 15D 0 LPI-CKV-OO-CV58 .
HPI-MOV-FT-1115E LPI-CKV-OO-CV50
LPR-MOV-FT- 1860A MSS-SRV-OO-SGSRV
LPR-MOV-FT-1860B .
HPI-MOV-FT-1350
LPR-MOV-Fr-1862A OEP-DGN-FS-DG01
LPR-MOV-FT-1862B CPC-MDP-FR-SWA24
LPR-MOV-FT-1890B CPC-CKV-OO-CV113
LPR-MOV-FT-1890A LPI-MDP-FR-B21HR
LPI-MDP-FR-A21HR
2nd order of OEP-DGN-FR-6HDG1
ACC-CKV-FT-CV128 magnitude
ACC-CKV-FT-CV130 0
OEP-DN-FS-DG03 0
ACC-CKV-FT-CV145 OEP-DGN-FS-DG02 0
ACC-CKV-FT-CV147 OEP-DGN-FR-6HDG2 0
HPl-MOV-FT-l 115B 0 SIS-ACT-FA-SISB 0
HPI-MOV-FT-1 l 15C 0
SIS-ACT-FA-SISA S
HI-MOV-FT-ll15D *:
OEP-DGN-FR-6HD3 0
HPI-MOV-FT-l lSE 0 HPI-CKV-FT-CV225
HPI-CKV-OO-CV258 . HPI-CKV-FT-CV410
HPI-MDP-FR-1A24H HPI-CKV-FT-CV25
LPR-MOV-FT-1860A RMT-ACT-FA-RMTSB
LPR-MOV-FT-1860B .0
RMT-ACT-FA-RMTSA a
LPR-MOV-FT-1862A .
AFW-TDP-FR-2P6HR 0
LPR-MOV-FT-1862B ACC-CKV-FT-CV147 S
LPR-MOV-FT-1890A 0 ACC-CKV-FT-CV145 0
LPR-MOV-FT-i890B ACC-CKV-FT-CV130
AFW-CKV-O00CV142 ACC-CKV-FT-CV128 S
AFW-TDP-FS-FW2 AFW-ACT-FA-PMP3B 0
HPI-MDP-FR-1A6HR AFW-ACT-FA-PMP3A 0
AFW-CKV-OO-CV157 HPI-MOV-FT-1867D
AFW-CKV-OO-CV172
AFW-MDP-FS-FW3A
AFW-MDP-FS-FW3B
HPI-MOV-FT-1350
SIS-ACT-FA-SISA
. _ ..
92
TABLE 20. ABSOLUTE ORDERING OF THE COMPONENT
CONTRIBUTORS: COMBINATION OF THE CASES
Contributor Above IE-3 Contributor Above 1E-5
HPI-CKV-FT-CV225 LPR-MOV-FT-1862A
HPI-MOV-FT-1115B
Contributor Above 1E4 HPI-MOV-FT- 11 15C
HPI-MOV-FT-1115D
ACC-CKV-FT-CV147 HPI-MOV-FT-1115E
ACC-CKV-FT-CV145 LPR-MOV-FT-1860A
ACC-CKV-FT-CV130 LPR-MOV-FT-1860B
ACC-CKV-FT-CV128 LPR-MOV-FT-1862A
HPI-MOV-FT-1115E LPR-MOV-FT- 1862B
HPI-MOV-FT- 115D LPR-MOV-FT-1890A
HPI-MOV-FT-1115C LPR-MOV-FT-1890B
HPI-MOV-FT-1 1 15B LPI-MDP-FS-SIIB
HPI-MDP-FR-1A24H LPI-MDP-FS-SI1A
HPI-CKV-O-CV258 LPI-CKV-00-CV58
LPR-MOV-FT-1890B LPI-CKV-OO-CV5O
LPR-MOV-FT-1890A MSS-SRV-OO-SGSRV
LPR-MOV-FT-1862B HPI-MOV-FT-1350
LPR-MOV-FT-1862A OEP-DGN-FS-DG01
LPR-MOV-FT-1860B CPC-MDP-FR-SWA24
LPR-MOV-FT-1860A CPC-CKV-OO-CV113
AFW-TDP-FS-FW2 LPI-MDP-FR-B21HR
AFW-CKV-OO-CV142 LPI-MDP-FR-A21HR
HPI-MDP-FR-1A6HR OEP-DGN-FR-6HDG1
AFW-MDP-FS-FW3B OEP-DGN-FS-DG03
AFW-MDP-FS-FW3A OEP-DGN-FS-DG02
AFW-CKV-OO-CV172 OEP-DGN-FR-6HDG2
AFW-CKV-OO-CV157
93
6. IDENTIFICATION OF RISK-DIRECTED AGING MANAGEMENT
STRATEGIES
6.0 Innyduction
Once the aging contributors are prioritized with regards to their risk contributions, then
aging management strategies can be focused on those contributors which are assessed to
be relatively high. This chapter illustrates how such risk-directed aging management
strategies can be identified and can be evaluated. The prioritized aging contributors
obtained in the previous chapter will be used as a demonstration. In many cases, the risk-
directed aging strategies need not necessarily involve large resources since the aging
controls can be pinpointed on the relatively few, top ranked contributors to the core
damage frequency increase caused by aging. The risk-directed aging management
strategies which are investigated involve scheduled replacement of the high aging
contributors and implementation of more frequent or more effective surveillance tests on
these high contributors. Since there are relatively few high contributors which have been
identified, the additional resources involved would not necessarily be large.
Furthermore, the additional resources required could be trade-off against less frequent
testing on the risk-unimportant components. The last section of the chapter further
investigates the impacts of replacing operational tests with condition-monitoring tests.
6.1 Risk-Directed Aging Management Strategies
We shall consider the base case prioritization results obtained in the previous chapter.
The same approaches would be used for the other results obtained using the different data
bases. In the previous chapter, Tables 12A and 12B give the dominant aging contributors
which are obtained for the base case. These tables are reproduced as Tables 21A and
21B on the following two pages. These contributors constitute approximately 99% of the
increase in core damage frequency due to aging.
From Tables 21A and 21B, the top contributors consists of the motor operated valves
(MOVs) in the Emergency Core Cooling System (HPWR). There are 14 motor
operated valves involved. The base case testing program involved testing the valves
every 18 months and replacing the valves at failure. The first risk-directed aging
management strategy investigated involves replacing the valves every 5 years
94
PLANT A: BASE CASE, SINGLE CONTRIBUTIONS
Plant A.- Single Contributions
TIRGALEX Aging Rates
TIRGALEX Testing Efficiencies
Total AC: I .8E04 year
Test
Rank Component Name Sensitivity Aging Rate M1hF Interval Aql AC
Coefficient (fr/yr) (months) (months) (/ye=)
LPR-MOV-FT-1862A 1.5E-04 3.6E-06 167 30 2.6E-01 3.9E-05
2 LPR-MOV-FT-1860A 1.5E-04 3.6E.06 167 30 2.6E-01 3.9E-05
3 LPR-MOV-FT-189OA 1AE-04 3.6E-06 167 30 2.6E-01 3.5E-05
4 HPI-MOV-Fr-1350 6.7E-05 3.6E06 167 30 2.6E-01 1.7E-OS
5 LPR-MOV-FT-1862B 2.1E-05 3.6E-06 167 30 2.6E-01 S.4E-06
6 OEP-DGN-FS-DG01 3.8E-04 3.6E-06 72 4 1AE-02 53E-06
7 LPR-MOV-FT-1860B 2.OE-05 3.6E-06 167 30 2.6E-01 5.3E-06
8 OEP-DGN-FR-6HDG1 3AE-04 3.6E-06 72 4 1.E-02 4.8E-06
9 OEP-DGN-FS-DG03 2.OE44 3.6E-06 72 4 1.4E-02 2.8E-06
10 OEP-DGN-FS-DG02 2.OE-04 3.6E-06 72 4 1.4E-02 2.8E-06
'C
(I,
11 OEP-DGN-FR-6HD03 1.9E-04 3.6E-06 72 4 1AE-02 2.7E-06
12 OEP-DGN-FR-6HDG2 1.7E-04 3.6E-06 72 4 1.4E-02 2.5E-06
13 PPS-MOV-FT-1535 9.5E-06 3.6E-06 167 30 2.6E01 2.4E-06
14 HPI-CKV-Fr-CV225 2.1E-03 4.0E-9 720 11 4.8E44 1.7E46
15 HPI-CKV-FT-CV25 2.1E-03 4.OE-09 720 11 4.8E-04 1.7E06
16 HPI-CKV-FT-CV410 2.1E-03 4.OE-09 720 11 4.8E-04 1.7E-06
17 HPI-MOV-FT-1115C 5.7E-06 3.6E-06 167 30 2.6E-01 1.5E06
18 HPI-MOV-FT-1115D 5.7E-06 3.6E06 167 30 2.6E01 1.5E06
19 HPI-MOV-FT-1115B 5.7E 06 3.6E-06 167 30 2.6EO1 1.5E06
20 HPI-MOV-Fr-1115E 5.7E-06 3.6E-06 167 30 2.6E-01 1.5E-06
21 LPR-MOV-FT-1890B 4.5E06 3.6E06 167 30 2.6E41 1.2E06
22 PPS-MOV-FT-1536 3.4E-06 3.6E-06 167 30 2.6E-01 8.8E-07
23 HPI-MOV-FT-1867D 2.9E06 3.6E06 167 30 2.6E-01 7.5E07
24 OEP-DGN-FR-DG01 5.0E-05 3.6E-46 72 4 1AE-02 7.2E-07
25 SIS-ACT-FA-SISA 1.8E05 3.0E-07 720 6 1.8E-02 5AE47
PLANT A: BASE CASE, DOUBLE CONTRIBUTIONS
Plant A. Double Coiuibuious

llRGALEX Aging Raw
TIRGALEX Testing Effcicws
Total AC: 7.6E-04 ycu
Tcst _Te
Rnk Component Nam Sensitivity Aging Raw Wv F Interval hql COMP0etNma Aging Raw P Ierval Aq2 AC
Coefficient (ib/yr) (1len1is) (mos) _l/yr) (months) (1mnu) (lyea)
HPI-MOV-FT-111SB 1.E-03 361-06 167 30 2.01 HP-MOV-Fr-I11SD 3.6E06 167 30 2.6E-01 1.3E-04
2 HKI-MOV-Fr-1I1C L9E43 3606 167 30 21-01 HPI-MOV-F-1115E 3.6E-06 167 30 2A-01 1.3E04
3 LPR-MOV-FT-1890A 1.5E-03 36E-06 167 30 2601 LPR-MOV-Fr-18908 36E-06 167 30 2.6E-01 1.0E-04
4 LPR-MOV-F-186OA 1SE-03 3.606 167 30 2.6E-01 LPR-MOV-F-1860B 3.6E-06 167 30 2.6E-01 1.004
5 LPR-MOV-Fr-1862A 15E-03 3.6E-06 167 30 2.6E1 LPR-MOV-F-1860B 3.6E-06 167 30 2.6E-01 1.004
6 LPR-MOV-FT-l86A 1.5E-03 3.6E-06 167 30 26101 LPR-MOV-FT-1862B 3.6E-06 167 30 2.6E-01 1.E-04
7 LPR-MOV-Fr-1862A 1.5E-03 3.6E-06 167 30 2.1-01 LPR-MV-Fr-186 3.6E06 167 30 2.6E-01 1.01-04
8 SIS-ACT-FA-SISB 6.5E-03 3.0E-07 720 6 1.8E-02 SIS-ACT-FA-SISA 3.0E-07 720 6 1.8E-02 5.9E-06
9 RMT-ACr-FA-RTSA 1.5E3 30E-07 720 6 1.8E-02 RMF-AC-FA-RMSB 3.OE-07 720 6 1E-02 1.4E-06
10 OEP-DGN-FR-6HDG3 5.6E-03 3.1-06 72 4 1.4E-02 0EP-DGN-FR-6HDG1 3.6E-06 72 4 14-02 I.IE-06
11 OEP-DGN-FS-DG01 49E-03 3.6E-06 72 4 1.4E-02 OEP-DGN-FS-DG02 3.6E-06 72 4 1AE-02 9.9E-07
12 OEP-DGN-FS-DG01 4.9E-03 36E-06 72 4 1.4E.02 OEP-DGN-FS-D03 36E-06 72 4 1AE4-02 9.9E-07
13 OEP-DGN-FS-DG01 4.0E03 3E6-06 72 4 IAE-02 OEP-DGN-FR-6HDG2 31-06 72 4 14-02 8.1E07
14 OEP-DGN-FS-DG01 4.0Ei43 3.6E-06 72 4 1AE-02 OEP-DGN-FR-6HDG3 3.1-06 72 4 1.E02 8.1E-07
15 OEP-DGN-FS-DG02 4.0F1-03 3.6E-06 72 4 1.4E402 0EP-DGN-FR-6HDCI 3.6E-06 72 4 IAE-02 8.1E-07
16 0EP-DN-FR-6HDGI 41-03 36E-06 72 4 1.4E-02 OEP-N-R 2 3.6E-06 72 4 1AE4-02 8.1E-07
17 OE.P-DGN-FS-DG 3 3.9E-03 3.6E-06 72 4 1.4E02 OEP-DGN-FR-6HDG1 3.6E-06 72 4 AE-02 7.8E-07
18 OEP-DN-FS-DG0 SE-03 3.6E-06 72 4 IAE4-02 MSS-SRV-00-SGSRV 7.01-07 22 22 3.41-03 2AE4-07
19 LPI-MDP-FS-SIlB 1.5E-03 2E-07 6 2 5.83-04 LPR.MOV-FT-1862A 3.63-06 167 30 2.6E-01 233E-07
20 LPl-MDP-FS-SllA 15E-03 2.01-07 86 2 51-04 LPR-MOV-F-1860B 3.6E-06 167 30 2.63-01 233E-07
21 LPI-MDP-FS-SllB 1.5E-03 20-07 86 2 5.8-04 LPR-MV-FF-1&0A 3.6E-06 167 30 2.6E-01 23E-07
22 LPI-MDP-FS-SllA I.5E43 2.0E-07 86 2 58-04 LPR-MOV-FI-1862B 3.6E-06 167 30 2.6E-01 23E-07
23 0P-DN-FR-6HDGI 4.6E-03 3.6E-06 72 4 1.4E-02 MSS-SRV-00-SOSRV 7E-07 22 22 3.4E-03 2.2E-07
24 PPS-MOV-FC.1536 2.9E-06 3.-06 167 30 2.63-01 PPS-MOV.FC-1535 3.6E06 167 30 2.6E-01 1.9E-07
25 OEP-DGN-FS-DG03 9.1E-04 31E-06 72 4 1.4E-02 0EP-DGN-FR-DG01 3.6E-06 72 4 IAE-02 1.8E07
(60 months) and improving the test and increasing the test frequency so that the effective
test interval is 6 months.
Tables 22A and 22B give the new results for this first risk-directed aging management
strategy. This enhanced aging management strategy results in an average core damage
frequency increase due to aging of 3.8 x 10-5 + 1.9 x 10- = 5.7 x 10-5 per year. This
compares to the base case core damage frequency increase of 1.8 x 10-4 + 7.6 x 104 =
9.4 x 10 4 per year. The risk-directed aging management strategy thus results in a factor
of 16 reduction in the core damage frequency increase due to aging. A significant
reduction in risk effects due to aging is thus achieved by focusing on the 14 risk-
important motor operated valves.
In addition to the motor operated valves, the prioritized aging contributors for the base
case in Tables 21A and 21B also identify as important risk contributors the diesel
generators (DGN), the three check valves (CKV) in the High Pressure Injection (HPI)
System, and the four Actuation Trains (ACr). Thus, a second risk-directed aging
management strategy is identified which involves carrying out the MOV tests and
replacements in the first strategy plus improving the test efficiency on the three diesels so
that the effective test interval is 1 month, replacing the three check valves every 20 years
(240 months) and replacing the four actuation trains every 20 years. (The necessary
requirements to improve the diesel test efficiency was assessed by personnel
knowledgeable of the test.)
Tables 23A and 23B give the new values for the aging contributors corresponding to the
second risk-directed aging management strategy. This second enhanced aging
management strategy results in an average core damage frequency increase from aging of
1.8 x 10-5 + 5.1 x 106 = 2.3 x 10-5 per year, which represents a factor of 41 reduction as
compared to the base case. Thus, by focusing on relatively few additional, risk important
aging contributors the core damage frequency increase due to aging is further
significantly reduced.
As a summary of strategies evaluated, Figure 7 illustrates the core damage frequency

increase due to aging for the base case, the first risk-directed aging management strategy,
and the second risk-directed aging management strategy. Instead of the absolute
increases, the relative core damage frequency increases due to aging may be of more
interest where the relative increase is the absolute increase divided by the original core
97
PLANT A: FIRST RISK-DIRECTED STRATEGY, SINGLE CONTRIBUTIONS
Plant A. Single Contributions
First Aging Control Strategy
TIRGALEX Testing Efficiencies Total AC: 3.8E-05 /year
Test
Ra& Component Name Sensitivity Aging Rate MTBF interval Aql AC
Coefficient ([ulyr) (months) (mont) (/Yea)
1OEP-DGN-FS-DG01 3.E-04 3.6E-06 72 4 1.4E-02 5.3E-06
2 OEP-DGN-FR-6HDG1 3.4E04 3.6E-06 72 4 1.4E-02 4.8E-06
3 LPR-MOV-Fr-1862A 1.5E04 3.6E-06 60 * 6 1I.9E-02 2.9E-06
4 LPR-MOV-Fr-1860A 1.SE-04 3.6E-06 60 * 6 1l.9E-02 2.9E-06
7 OEP-DGN-FR-6HDG3 1.9E404 3.6E-06 72 4 1.4E42 2.7E06
8 LPR-MOV-FT-1890A IAE-04 3.6E-06 60 * 6 I.9E-02
1 2.6E-06
9 OEP-DGN-FR-6HDG2 1.7E-04 3.6E-06 72 4 1.4E-02 2.5E-06
10 HPI-CKV-FT-CV225 2.1E-03 4.OE-09 720 11 4.8E-04 1.7E-06
11 HPI-CKV-Fr-CV25 2.1E03 4.OE-09 720 11 4.8E-04 1.7E-06
ao 12 HPI-CKV-Fr-CV410 2.1E-03 4.OE-09 720 11 4.8E-04 1.7E-06
13 HPI-MOV-FT-1350 6.7E-05 3.6E06 60 * 6 * 1.9E-02 1.3E-06
14 OEP-DGN-FR-DG01 5.0E05 3.6E-06 72 4 1.4E-02 7.2E-07
15 SIS-ACT-FA-SISA 1.8E-5 3.OE-07 720 6 1.8E-02 5.4E-07
16 LPR-MOV-Fr-1862B 2.1E-05 3.6E06 60 * 6 1l.9E-02 4.OE-07
17 LPR-MOV-Fr-1860B 2.OE-05 3.6E-06 60 * 6 * 1.9E-02 3.9E-07
18 PPS-MOV-FT-1535 9.5E-06 3.6E-06 60 * 6 * 1.9E-02 1.8E-07
19 HPI-MOV-FT-11158 5.7E-06 3.6E-06 60 * 6 * l.9E.02 I.IE-07
20 HPI-MOV-FT-ll15D 5.7E06 3.6E-06 60 ' 6 * 1.9E-02 l.E-07
21 HPI-MOV-Fr-1115C 5.7E-06 3.6E-06 60 * 6 * 1.9E-02 IE-07
22 HPI-MOV-Fr-1115E 5.7E-06 3.6E-06 60 * 6 1.9E-02 .1E-07
23 LPR-MOV-FT-1890B 4.5E-06 3.6E-06 60 * 6 * 1.9E-02 8.5E-08
24 PPS-MOV-FT-1536 3.4E-06 3.6E06 60 * 6 * 1.9E-02 6.5E-08
25 HPI-MOV-FI-1867D 2.9E06 3.6E-06 60 6 * 1.9E-02 5.6E-08
* - Control Value
PLANT A: FIRST RISK-DIRECTED STRATEGY: DOUBLE CONTRIBUTIONS
PlantA: Double Contributions

Second Aging Control Stategy
TIRGALEX Testing Efficiencies Total AC: 5.1E06 year
Test Test
Rank Component Nae Sensitivity Aging Rate MTBP Interval &qi Component Name Aging Rate MTDP Interval Aq2 Ac
Coefficient (/hrlyr) (mnths) (mot) (or/yf) (months) (months) Vyeff)
I HPI-MOV-FT-111SC 1.9E603 3.E.06 60 6 1.9-2 HPI-MOV-FT-1115 3.6E-06 60 6 1.902 6.8E-07
2 H1P-MOV-FI-11SB 1.98E03 3A-06 60 6 1.9E-02 HPI-MOV-Fr-1115D 3.6-06 60 6 1.9E.02 6.SE-07
3 LPR-MOV-FT-1862A 103 3.66h06 60 6 1.9f.02 LPR-MOV-Fr-1862B 3.6E-06 60 6 I.9E02 S.4E.07
4 UR-MOV-FT-186OA I.5-03 3.6E-06 60 6 19E-02 LPR-MOV-FT-1860B 3.6E06 60 6 I.9E-02 S.4E47
S LPR-MOV-FT-1860A 156-03 3.6E606 60 6 I.9E-02 LFR-MOV-Fr-1862B 3.6E06 60 6 1.9E02 5.4E-07
6 UR-MOV-FT-1890A 1S03 3.6E-06 60 6 I.9E-02 UR-MOV-FT-1890B 3.6E-06 60 6 19E-02 5AE-07
7 LPR-MOV-FT-1862A 15E-03 3.6E06 60 6 I.9E-02 LPR-MOV-Fr-18608 3.6E-06 60 6 I.9E-02 5.4E.07
8 SIS-ACr-FA-SISB 6.5E-03 3.0E-07 240 6 6.0E-03 SIS-ACT-FA-SISA 3.0E-07 240 * 6 6.OE-03 2.3E-07
9 0EP-DGN-FR-6HDG3 S.6E-03 3.6E-06 72 1* 3.9E-03 OEP-DN-FR-a I 3.6E-06 72 1 3.9E-03 8.6E0
10 OEP-DN-FSDGO01 4.96-03 3.-06 72 1 3.9E-03 OEP-DGN-FS-DG03 3.6E-06 72 1 * 3.9E-03 75E-08
11 OEP-DN-FS-WD01 4.9-03 3.6E-06 72 1 3.9E03 0EP-DN-FS-DO2 3.6E4-6 72 1 * 3.9E-03 7.5E-08
12 0EP-DGN-FS-DG01 5.6-03 3.6E-06 72 1 * 3.9E-03 MSS-SRV-OSGSRV 7.0E-07 22 22 3AE-03 6.7Ea
13 OEP-DGN-FS-DG01 4.0643 3.6E-06 72 * 3.9E-03 OEP-DGN-FR-6HD02 3.6E606 72 1a 3.9E03 62E08
14 0EP-DN-FS-DW01 4.0E-03 3.6E-06 72 1 * 39E-03 0EP-DGN-FR-6HDG3 3.6E-06 72 1 * 3.9E-03 6.2E08
15 0EP-DN-FS-DO2 4.04-03 3.6E-06 72 1a 39E43 0EP-DN-FR-DGI 3.6E06 72 1* 3.9E03 6.2E08
16 OEP-DGN-FR-6HDG1 4.6E-03 3.6E06 72 1 * 39E03 MSS-SRV-00-SGSRV 7.0E-07 22 22 3AE-03 6.1E-08
17 OEP-DGN-FR-6HDG1 4.0E-03 3.6E-06 72 1 O.3.9E-03 OEP-DGN-FR6HDG2 3.6E-06 72 1 3.9E-03 6.1E08
18 OEP-DGN-FS-DG03 39E-03 3.6E-06 72 1 39E-03 0P-DGN-FR-6HDGI 3.6E-06 72 1 a 3.9E-03 6.OE-08
19 RMT-ACr-FA-RMTSA 15-03 3.0607 240 a 6 6.0E03 RMTACr-PA-RMSB 3.0-07 240 a 6 6.0E-03 SAE08
20 LPI-MDP-FS-SIIA IS6-03 2.0-07 86 2 5.86-04 LPR-MOV-Fr.1862B 3.6E-06 60 6 196-02 1.7E08
21 LPI-MDP-PS-SIIB 1.5E-03 2.0E07 86 2 5.8E.04 LPR-MOV-Fr-1860A 3.6E-06 60 6 I.9E-02 1.7E-08
22 LPI-MDP-FS-SIIB 15E.03 2.067 86 2 5.8-04 LPR-MOV-F-1862A 3.6E-06 60 6 19E-02 1.7E48
23 LPI-MDP-FS-SIIA 15-03 2.0E-07 86 2 5.804 LR-MOV-Fr-l86R 3.6E-06 60 6 19E-02 1.7E.08
24 OEP-DGN-FS-DG03 9.1-04 3.6E-06 72 1 3.9E-03 0EP-DON-FR-DGO01 3.6E-06 72 1 * 39E-03 1AE-08
25 PPS-MOV-FC-1536 2.9E-6 3.6E-06 60 6 I.9E-02 PPS-MOV-FC-1535 3.6£-06 60 6 1.9E-02 .IE-09
*- CotlAtO VAIe
PLANT A: SECOND RISK-DIRECTED STRATEGY, SINGLE CONTRIBUTIONS
Plant A. Single Contributions
Second Aging Control Strategy
nRG AEXTesting Efficiencies Total AC; .8E-05 /year
Test
Rank Component Name Sensitivity Aging Rate MTF Interval Aq1 AC
Coefficient (/hr/yr) (months) (months) (/year)
1 LPR-MOV-FT-1862A 1.5E-04 3.6E-06 60 6 I.9E.02 2.9E-06
2 LPR-MOV-FT-1860A 1.5E-04 3.6E 06 60 6 1.9E-02 2.9E.06
3 LPR-MOV-Fr-1890A IAE-04 3.6E-06 60 6 1.9E-02 2.6E-06
4 OEP-DGN-FS-DG01 3.8E-04 3.6E-06 72 1 * 3.9E-03 1.5E-06
5 OEP-DGN-FR-6HDGl 3AE-04 3.6E-06 72 1 * 3.9E-03 13E-06
6 HPI-MOV-FT-1350 6.7E.05 3.6E 06 60 6 1.9E-02 13E-06
7 OEP-DGN-FS-DG02 2.OE04 3.6E-06 72 1 * 3.9E-03 7.8E07
8 OEP-DGN-FS-DG03 2.OE-04 3.6E-06 72 1 3.9E03 7.8E-07
9 OEP-DGN-FR-6HDG3 1.9E-04 3.6E-06 72 1 * 3.9E03 7.5E-07
10 OEP-DGN-FR-6HDG2 1.7E-04 3.6E.06 72 1 * 3.9E03 6.8E-07
11 LPR-MOV-F1-1862B 2.1E-45 3.6E-06 60 6 1.9E.02 4.OE-07
12 LPR-MOV-Fr-1860B 2.OE-05 3.6E-06 60 6 1.9E-02 3.9E-07
13 HPi-CKV-Fr-CV225 2.1E-03 4.OE-09 240 * 11 1.6E.04 3.4E-07
14 RPI-CKV-Fr-CV4l0 2.1E-03 4.OE-09 240 * 11 1.6E-04 3.3E-07
15 HPI-CKV-FT-CV25 2.IE-03 4.0E-09 240" 11 1.6E-04 33E-07
16 OEP-DGN-ER-DG01 5.OE-05 3.6E-06 72 1 * 3.9E-03 2.OE47
17 PPS-MOV-FT.1535 9.5E-06 3.6E 06 60 6 1.9E-02 1.8E-07
18 HPI-MOV-FT-1115B 5.7E-06 3.6E-06 60 6 1.9E 02 l.lE-07
19 HPI-MOV-FT-II1SE 5.7E-06 3.6E-06 60 6 I.9E02 I.IE-07
20 HPI-MOV-Fr-1115D 5.7E.06 3.6E-06 60 6 1.9E02 l.E07
21 HPI-MOV-Fr-1115C 5.7E-46 3.6E-06 60 6 1.9E-02 1.1E-07
22 SIS-ACT-FA-SISA 1.8E-05 3.0E-07 240 6 6.OE.03 L.IE-07
23 LPR-MOV-Fr-1890B 4.5E-06 3.6E-06 60 6 1.9E02 8.5E.08
24 PPS-MOV-FT-1536 3.4E-06 3.6E-06 60 6 1.9E02 6.5E-08
25 HPI-MOV-Fr-1867D 2.9E06 3.6E-06 60 6 I.9E.02 5.6E48
* - Control Value
PLANT A: SECOND RISK-DIRECTED STRATEGY, DOUBLE CONTRIBUTIONS
Plant A: Double Contribution

Fust Aging Control Strategy
TRGALEX Aging Rates
TIRGALEX Testing Efficiencies Total, C: 1.9E-05 yet
Test Test
Rank ComponentNane Senitiviy Aging Rate MTBP Inteval Aqi Component Name Aging Rate WMFB Interval Aq2 AC
Coefficient _h/'yr) (monds) (months) (flh/yr) (months) (months) Qbw)
I SIS-ACT-FA-SISB 6.5E403 3.0E7 720 6 1.8H-02 SIS-ACr-FA-SISA 3.0=-07 720 6 l.86.02 5.96.06
2 RMT-ACT-FA-RMTSA 1.5E-03 3.4-07 720 6 1.8E02 RMI-ACr-FARMTSB 3.OE07 720 6 1.8E-02 1AE-06
3 OEP-DGN-FR-6D03 5.6E03 3.6-06 72 4 IAE4Q OEP-DGN-FR-6HDGI 3.6E06 72 4 I4E-02 I.IE-06
4 0EP-DGN-FS-D01 4.03 3.6E-06 72 4 1.4E02 OEP-DGN-FS-D02 3.6E-06 72 4 1AE-02 9.9E-07
S OEP-D0N-FS-DG01 4.9E43 3.6E06 72 4 1.4E-02 OEP-DGN-FS-DG03 3.6E06 72 4 1AE402 9.9E07
6 0EP-DGN-FS-DG01 4.0E-03 36E-06 72 4 IAE-02 OEP-DGN-FR-6HDG2 3.6E06 72 4 1.4E-02 8.1E-07
7 OEP-DGN-FS-D01 4.0E03 3.6E-06 72 4 IAE-02 OEP-DGN-FR-6G3 3.6E06 72 4 1AE02 8.1E07
8 OEP-DGN-FS-D002 4.0E03 3.6E06 72 4 1.4E-02 OEP-DGN-FR-6HDGI 3.6E-06 72 4 14E02 8.1E07
9 OEP-DGN-FR-6HDG1 4.0E-03 3.6E06 72 4 1.E02 OEP-DGN-FR-6HDG2 3.6E-06 72 4 1.4E-02 8.1E07
10 OEP-N-FS-DG03 3.9E03 3.6E-06 72 4 E4-02 OEP-DGN-FR-6HDGI 3.6E-06 72 4 1.4E-02 7.8-07
C0
11 HPl-MOV-FT-1115B .9E43 3.6E-06 60 * 6 .9E-02 HPI-MOV-Fr-1115D 3.6E-06 60 * 6 * I9E02 6.8E-07
12 HPI-MOV-FT-111SC I.9E-03 3.6E-06 60 * 6 I9E-02 HPI-MOV-Fr-1115E 3.6E.06 60 * 6 * .9E-02 6.8-07
13 LPR-MOV-Fr-1862A I SE-03 3.6E-06 60 * 6 * 19E.02 LPR-MOV-FF-1860B 3.6E06 60 * 6 * I9E-02 5.4E-07
14 LPR-MOV-FT-1862A IE-03 3.6E-06 60 * 6 1.9E02 LPR-MOV-FT-1862B 3.6E06 60 * 6 * I.9E02 5.46.07
15 LPR-MOV-FT-1860A 1.5E03 3.6E06 60 * 6 1 9E02 LPR-MOV-Fr-1862B 3.6E06 60 * 6 0 I.9E-02 5.4E.07
16 LPR-MOV-Ft-1860A 1.5E-03 3.6E-06 60 * 6 * .9E-02 LPR-MOV-FT-1860B 3.6E-06 60 0 6 * 19E-02 5.4.07
17 LPR-MOV-F-1890A 1.5-03 3.6E-C6 60 * 6 19E02 LPR-MOV.FT-18M0B 3.6E-06 60 ^ 6 0 1.9E-02 S.4E-07
18 OEP-DGN-FS-DG01 S.OE-03 3.6E-06 72 4 1.4E-02 MSS-SRV-SGSRV 7.06-07 22 22 3AE03 2.4E-07
19 OEP-WN-FR.G1 4.6E-03 3.6E-06 72 4 1.4642 MSS-SRV-00-SGSRV 7.0E-07 22 22 3.4E-03 2.2E-07
20 OEP-DGN-FS-D03 9.1E44 3.6E-06 72 4 1A4-02 OEP-DGN-FR-DG01 3.6E-06 72 4 IAE-02 1.8E-07
21 LPI-MDP-FS.SIIA 1.5E-03 2.0647 86 2 5.8E-04 LPR-MOV-FT-1862B 3.6E-06 60 * 6 * 1SE-02 1.7E-08
22 LPl-MDP-FS-S1B ISE-03 2.0E07 86 2 5.8E-04 LPR-MOV-FT.1862A 3.6E-06 60 * 6 * I9E-02 1.7E8
23 LPI-MDP-FS-SlB 1.6-03 2.OE-07 86 2 5.8E44 LPR-MOV-Fr-1860A 3.6E-06 60 * 6 l.9E-02 1.7E-08
24 LPI-MDP-FS-SlA 1.5E-3 2.0E-07 86 2 5.8-04 LPR-MOV-FT-1860B 3.6E-06 60 * 6 * 1.9E02 1.7E-8
25 PPS-MOV-FC-1536 2.9E06 3.6E06 60 * 6 19E-02 PPS-MOV-FC-1535 3.6E-06 60 * 6 * 19E-02 .IE-09
- conuol Value
FIGURE 7. CORE DAMAGE FREQUENCY INCREASE DUE TO AGING:
MINIMAL MAINTENANCE VERSUS RISK-DIRECTED CONTROL
1.OE-03: ............................... ......................................................

......................................................................... ..................
.............................................................. ............................
AC
per year
1.OE-04. ..............................................................................
:::::::::::::::::::::::::::':::::::::::::::::::::::::::::::::::::::::::.:::::::::::.............::
. ....... ..
. ... . . ............... . .... . .. . 0...
................ ... . . . . . . .. . . . .. . . . .. .. . . . . . .. . .. .. .. . . . . .
oowoowoow
ow
ow M1lm
................... 11
~l l ll l T11.............................
,tXX,/X - /,zZ, 111 m I...........m.........1......
111.....
1.OE-05
PWR
_ MINIMAL MAINTENANCE M RISK-DIRECTED STRATEGY I

EE RISK-DIRECTED STRATEGY 2
FIGURE 8. RELATIVE CORE DAMAGE FREQUENCY INCREASE DUE TO

AGING: MINIMAL MAINTENANCE VERSUS RISK-DIRECTED CONTROL
1.OE+02 : ~~~~~ ~ ~ ~ ~ ~~ ...............

..................................................
..... =:.:=.................................
:............................................................................................
.......
.......................
........................ ........ ....... ............................. .......................... ................. ............... ......................................
..... ..................................................... ......................................................... ............ ........... ................. ..................
................................
..
AC ......... ..... .m !...... ...................................................................................................
................... .... |||S.......

............................................................................. ...............
per year
1.OE+01
::.::: . . . .. . ... . .. . :: -:-::
. ;;......
:::.::; ;..;;. . . . . . . . . . . .. . . . . . . . . . . . .
...
..
...... ............. ............ ..
.. .... ........................................... ............................
...
..................... .................. ............................. ............................ ..
,.......................... ..... .................. ......................................... .......... . ...
........................ ..................... ................................................................
................................................................................ ............................................................
1.OE+OO:
:,nr. .
1.OE-01 -
PWR
_ MINIMAL MAINTENANCE M RISK-DIRECTED STRATEGY 1

M RISK-DIRECTED STRATEGY 2
102
damage frequency increase in the standard PSA. Figure 8 illustrates the relative core
damage frequency increase due to aging for the base case and the risk-directed aging
management strategies. ( The tables of aging contributions can be simply changed to
relative contributions by dividing by the original PSA core damage frequency). Figures
7 and 8 again illustrate the significant control of aging effects which can be achieved by
focusing on the risk important aging contributors.
Other alternative aging management strategies directed toward the risk important
contributors could be evaluated by using the appropriate test and maintenance models to
calculate the aging effects Aq and the subsequent core damage frequency impact AC.
The core damage importances Si, Sij would remain the same and would not need to be
recalculated. The formulas and approaches which have been presented allow these
alternative evaluations to be straightfowardly carried out.
103
7. TIME DEPENDENT AGING EVALUATIONS
7.0 Intducion
In evaluating the risk effects of aging, either the average effects of aging or the time
dependent effects of aging can be calculated. The previous two chapters focused on
evaluations of the average risk effects of aging. Determining the average effects of aging
involves averaging the time dependent component unavailabilities over the period
between replacements of the component or over given time periods (such as the plant
lifetime). The average component unavailabilities or unavailability increases are then
used in the PSA model to determine the average increase in core damage frequency and
other average risk effects. These average aging contributions are straightforward to
calculate and generally have reduced uncertainties due to the time averaging effect.
When time dependent aging effects are determined then time dependent component
unavailabilities (or unreliabilities) are utilized. The time dependent component
unavailabilities are input to the PSA to obtain the time dependent core damage frequency
with aging or the time dependent increase in core damage frequency due to aging. The
average aging contribution can also be obtained by time averaging the results over
appropriate periods of time. The time dependent aging results provide more detailed
information, however they involve more detailed calculations and have larger
uncertainties due to the additional details.
This chapter demonstrates how time dependent aging results can be obtained using the
models and approaches given in the previous chapters. Basically, to obtain time
dependent aging results, the aging contributions at a given time are used in the formulas,
e.g. Equation (45), instead of the average aging contributions. The calculations are
repeated at different time points with the appropriate time dependent component
unavailabilities used at each time point. The core damage frequency which is obtained at
each time point then provides a time track of the effects of aging which can be used for
monitoring and predicting aging effects. However, more accurate component aging
failure rates are required for these time dependent evaluations.
104
7.1 Basic Time Dependent Eauations
We shall again consider the risk importance approach for the APSA approach, however
the general time dependent models we present apply to any APSA approach used. For
the risk importance approach, the increase in core damage frequency AC due to aging is
again from Equation (45) of Section 3.3
ac = Si~li + Sij.. qihqj+.--+YSI2-nAql-.. qn (63)

j>i
Si, Sij, etc. are again the core damage frequency importances determined from the
standard PSA. The increases in component unavailability Ach due to aging are now the
time dependent unavailability increases. Now, from standard reliability theory, and as.
also given in the appendix, the time dependent unavailability q(t) at given time t is
generally
q(t) = 1-exp(- JA(w)dw) 4)

(tT-tR)
where
X(w) the age dependent component failure rate at age w (65)
tR = the time of the last replacement or renewal of the component (66)
and
tT - the time of the last operational surveillance test (good as old (67)
test) of the component
the symbol.(tT - tR)+ denotes the maximum of zero and (tT - tR), i.e.
(tT-tR) =tT-tR; tT>tR (68)
105
=0 ;tTStR (69)
Thus, in Equation (64) if the last renewal is before the last test (tT>tR), the age of the
component at the start of the test is tT - tR. If the renewal occurs after the test (tTtR)
then the age starts again at zero.
The time dependent increase in component unavailability Aq due to aging is the

difference between the time dependent component unavailability with aging q(t) and the
time dependent unavailability without aging q,(t)
Aq= q(t)-qO(t) (70)
where
qO(t) = l-ex(-A(t-tT)) (71)
and where X is the constant component failure rate. Note that the last time of renewal tR
does not enter q,(t) since aging and hence renewal is not considered. The average value
of %(t) over a test interval is the average component unavailability used in a standard
PSA. The average value of Aq(t) over a renewal interval is the average increase in
component unavailability due to aging which was utilized in the applications discussed in
the previous two chapters along with the assumption of a linear aging failure rate.
7.2 Time Dependent Equations for a Linear Aging Failure Rate
We will consider the linear aging failure rate with no threshold and will describe how the
threshold age can be later incorporated into the results. The linear aging failure rate with
no threshold is again
A(w)=A+aw (72)
where X is the constant failure rate, a is the aging rate and w is the age of the component.
Substituting the linear aging failure rate into the general formula for q(t), Equation (64),
gives
106
t-tft 1 (73)
q(t)=1-exp - J(A +aw)dw
(tT-tR) .
We shall consider the case when tR<tT and hence (tT - tR)+ = tT - tR. When tR2tT then
the results which are obtained will apply by assuming tT=tR since the time of last renewal
also then serves as a the time of the last equivalent surveillance test. For tR<tT
t-tR
q(t)= 1-exp - |(A+aw)dw (74)
tT-tR
or
q(t) = 1-exp{-4(t- tT) a((t - tR)2 -(tT -tR)2)] (75)
The difference in availability Aq(t) is then
Aq(t) = 1-exp[)A(t-tT)-2a((t-tR) -(tT - tR) )]-[1-ekp(-A(t-tT))] (76)
Expanding the exponential to first order gives the simple result
Aq(t)= 1 a((t-t)2-(tT-tR))
2
This first order equation gives accurate results to at least two significant figures when
Aq(t) < 0.1 otherwise, the exact equation, Equation (76), can be used. A threshold age
can be incorporated into Equation (76) or (77) by subtracting the threshold age from t;
the squared terms in the Equations(76) and (77) will then only be nonzero if they are
larger then the threshold age.
If the component is tested at intervals of T and is replaced or renewed at intervals of L

then
107
tT = []T (78)
and
n tR -(79)
ILI
where "[x]" denotes the greatest integer function, e.g. [9.34] = 9. For periodic testing
and renewal, Equations (78) and (79) can then be substituted into Equations (76) or (77)
for Aq(t).
7.3 Applications of the Time Dependent. Linear Aging Failure Rate Equations
For a given application, Aq(t) can be determined at a given time point for each aging
component using Equation (76) or (77). The Aq(t) for each aging component can then be
substituted into Equation (63) for the time dependent core damage frequency due to
aging AC = AC(t). These calculations are then repeated for the time points of interest.
Note that the core damage frequency Si, Sij ,etc. are not dependent on time and need to
be computed only once from the standard PSA.
Figure 9 shows the time dependent core damage frequency increase AC(t) due to aging
that is calculated for the basic case evaluation carried but in Chapter 5. The same base
case data is used as in Chapter 5, except that time dependent aging effects are calculated
instead of time-averaged aging effects. As in Chapter 5, the renewal interval for each
component is assumed to be the components mean time to failure. In Figure 9, the time
points are selected to correspond to the test times for each component. The components
with the same test interval are assumed to be tested at the same time, sequentially so as
not to violate technical specifications.
A more general evaluation of Aq(t) can be obtained by using the equations in Appendix
which can account for nonlinear aging effects and for the randomness of failure times
when components are only replaced at failure. Figure 9 is useful for showing the time
tracking information which is obtained from time dependent evaluations. The total core
damage frequency at any given plant age is then AC(t) plus the baseline value in the
108
FIGURE 9. TIME DEPENDENT CORE DAMAGE FREQUENCY INCREASE AC
DUE TO AGING
3.5E-03
Actives
(per year)
2.5E-03 . -- .. . . .
2.OE-03 . .. . .. /
OI .5E-03
5.OE-04 /t; /f~~~~~~~~~~~~~~-

atses
5 .OEE 4 . ..... .......................... ... ... I.... ......................
O.OE+00
0 100 200 300 400 500
Plant Age (Months)
PSA. The sudden drops in the core damage frequency increase AC(t) in Figure 9 occur
when the motor operated valves in the Emergency Core Cooling System are replaced at
failure. The high peaks in AC indicate that replacement only at failure for these critical
components is not adequate, and more frequent replacement or overhaul of the dominant
failure contributors is needed. This is a similar result as found in Chapter 5 where
average aging effects were utilized. Calculations of the time dependent aging effects.
provides useful information on the growth of the aging effects, indicating the times at
which the aging effects become significant
110
8. SENSITIVITY AND UNCERTAINTY EVALUATIONS
8.0 Intmduction
One of the most useful applications of an APSA is for sensitivity and uncertainty studies.
Both sensitivity studies and uncertainty analyses involve systematically changing
variables in the APSA and then determining the core damage frequency changes. An
uncertainty analysis differs from a sensitivity study in that probability distributions are
also assigned to the variables which represent the probabilities for the different values
which a variable may assume. Using uncertainty propagation techniques, a probability
distribution is then determined for the core damage frequency, which gives the
probabilities for the different core damage frequency values. A sensitivity study does not
utilize probability distributions but simply changes the input values and determines the
resulting change in the core damage frequency.
The variables which are changed in a sensitivity study or uncertainty study can not only
be data values, but can also be models and assumptions. In an APSA, sensitivity or
uncertainty studies can be performed for different possible aging behaviors for the
components, for alternative maintenance effects, alternative test effects, or alternative
repair effects. By designing appropriate sensitivity or uncertainty studies, the capability
of a given test and maintenance program to control aging risk effects over a range of
plausible alternatives can be systematically evaluated. These evaluations can help guide
and focus aging management activities.
This chapter illustrates uncertainty and sensitivity analyses which can be carried out
using an APSA. The following section first shows an application of uncertainty analysis
by assigning uncertainty distributions to the data used in an APSA. The subsequent
sections illustrate sensitivity analyses which can be carried out to evaluate the risk
sensitivities to aging and to maintenance strategies.
8.1 Uncertainty Analysis of Data Used in an APSA
The uncertainty analysis which is described is also described in NUREG-1362 (9).

Further information is provided here on the basic principles and applications of such an
analysis. An uncertainty analysis is carried out in an APSA to assess the uncertainties
one has in the aging results and to identify the dominant contributors to the uncertainty.
111
A data uncertainty analysis involves assigning error ranges to the data used in an APSA.
The error range assigned to a data input represents the different possible values the data
input may have because of uncertainties associated with collecting and estimating the
data. The probability distribution associated with the error range gives the probability
that the data will have any of the specific values in the range.
In a standard PSA, the input data which are assigned associated uncertainties include the
initiating event frequencies, component failure rates, test and maintenance intervals and
downtimes, and human error rates. For an APSA, the additional input data which have
uncertainties include the component aging failure rates, and test and maintenance data
describing the aging control effects of the test or maintenance. The component aging
failure rates will generally have the largest uncertainties.
To illustrate a data uncertainty analysis performed for an APSA, uncertainties are

determined for the aging impacts calculated in Chapter 5 which utilized the risk
importance approach and the linear aging failure rate model. Error ranges are assigned to
the data used for the base case evaluation. Error ranges are specifically assigned to the
component aging rates which are used, to the component mean times to failure (MTBF)
which are used as the component replacement times, and to the TIRGALEX testing
efficiencies which are used. Error ranges are also assigned to the risk importance
coefficients Si, Sij coming from the PSA.
An error ranges is generally defined in a PSA by assigning an error factor to a midpoint

data value. The upper value of the error range is defined as the data value times the error
factor and the lower value is the data value divided by the error factor. The data value is
thus the midpoint value of the range. This error range is generally defined to be a 90%
error range with the upper value being a 95% value and the lower value being a 5%
value. There is thus a 90% probability that the data value will lie in the defined range.
The error factors assigned to each piece of data for the APSA uncertainty analyses are
given below:
Data Type Error Factor
Component aging rate 10
Surveillance test efficiency 2
Component MTBF 2
Risk importance coefficient 5
112
The assignment of the above error factors is described in NUREG-1362 (9) and also in
NUREG/CR-5510 (7). Basically, the error factor of 10 on each component aging rate
covers uncertainties and plant specific variations that were observed by comparing the
TIRGALEX aging rate values with values estimated using plant failure data for selected
components. The error factor of 2 for each test efficiency basically gives test efficiency
values from 0 to 1 (the efficiencies are truncated at 1). The error factors for each
component MTBF and each risk importance coefficient (Si and Sjj) are assigned based on
PSA uncertainties.
The probability distribution assumed for each error range is a log uniform distribution
with the given upper and lower values. The log uniform distribution is a flat distribution
on a log scale. The log uniform is used in PSA uncertainty analyses when any factor
increase in the data value or any factor decrease in the value, within the range of possible
values, is equally likely. The log uniform distribution is thus a nonpreferential
distribution.
Before using the error ranges and probability distributions, one must decide whether
different data values are independent or are correlated with one another. Recent PSAs
assume data for a similar components to be totally correlated. (See for example
Reference (1).) For example, failure rates for all motor operated valves are assumed to be
totally correlated. When data are totally correlated then the data follow the same
probability distribution and are not independent with separate, similar distributions. To
be consistent with PSA approaches, all aging rates, MTBFs, and testing efficiencies for a
given type of component (eg. motor operated valves) are assumed to be totally
correlated. Further details are given in NUREG/CR-55 10.
The error ranges and probability distributions for the data are then propagated through
the APSA calculations to obtain the error ranges and probability distributions on the
APSA results. Standard Monte Carlo simulation approaches are used in the PSA for this
propagation and these same approaches can be applied to the APSA. References (1)
through (3), as well as NUREG/Cr-5510 (7), describe the simulation approaches and
computer codes which are available to automatically carry out the calculations once the
data error ranges and probability distributions are assigned.
Figure 10 shows the output of the uncertainty analyses in terms of the probability
distribution obtained for the core damage frequency (CDF) increase due to aging.
113
FIGURE 10. UNCERTAINTY ANALYSIS OF AGING EFFECTS: PLANT A
CCDF
100%
80%
60%
p"
40%
20%
0%
-6 -5.5 -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1
Log (Increase in Core Damage Frequency)

The y-axis of the figure is the probability that the CDF increase exceeds a given value on,
the x-axis. The x-axis is the log to the base 10 of the CDF increase; the x-axis is thus the
exponent of 10 of the CDF increase. The curve of the probability distribution on these
axes is termed a complementary cumulative distribution function, or ccdf, in standard
PSA terminology.
In the figure, ccdf curves are given for the single component contribution to the CDF
increase, for the double component interaction contribution, and for the total CDF
increase due to aging. The ccdf curves can be used to obtain error ranges for the output
results. For example, a 90% error range for the total CDF increase has a lower value of
approximately lx1O 3 5 = 3x104 per year (corresponding to a y-value of 95%) and an
upper value of approximately lxlIO-- 5 = 3x10-2 per year (corresponding to a y-value of
5%). More accurate values can be obtained from the associated tables which are
produced as part of the uncertainty analysis. Error ranges for the single and double
contributions can be obtained in a similar manner.
Uncertainty ccdfs and associated error ranges can be obtained for every aging
contributor, in addition to the sum totals for the singles and doubles shown in Figure 10.
These ccdfs and error ranges can be used to describe the uncertainties and confidences
which are associated with quantifying the CDF effects of aging. The lower end point of
an error range (eg. the lower bound) can be used to define the minimal aging effect and
the upper end point (the upper bound) can be used to define the maximum aging effect
which account for data uncertainties. The ccdf curves themselves can also be used in
more formal decision approaches for cost-benefit and optimization studies.
8.2 Sensitivity Studies of the Effects of Different Aging Rates
Sensitivity studies differ from uncertainty analyses in that no probability distribution is

assigned to the input variables which are changed. Only input values are changed. By
appropriately designing a sensitivity study and by systematically varying selected
variables, significant amount of information can be obtained on the sensitivities of risks
to aging effects and to test and maintenance practices.
To investigate the sensitivity of the core damage frequency effects to aging, a very
useful and straightforward sensitivity study to carry out is to vary component aging
failure rates over a plausible range and then determine the resulting core damage
115
frequency effects under given maintenance practices. To aid in the interpretation of the
results, the component aging failure rates can be expressed in a relative form which
indicates the relative size of the aging occurring. For the linear aging failure rate model,
the component aging rate is thus expressed as a relative percentage increase per year in
the baseline component failure rate.
If a is the component linear aging failure rate, then expressing a as a relative fraction of
the constant, baseline component failure rate Xb, we have
a rAO (80)
or
r= a (81)
10
In percentage (%) terms
r(in %) = a x 100 (82)

Ao
If the aging rate a has units of the failure rate change per year as in Table 8 in Chapter 5
then r has the units of per year. Thus, r gives the percentage increase in the component
failure rate per year due to aging. Note, that r can be larger than 100% per year.
Expressing the aging rate of the component as a relative percentage increase per year has
several advantages for aging sensitivity studies. The aging rate of a component can be
systematically varied by varying r from 0% through 100%. Higher values of r can also
be used. The aging rates of a group of components can also be simultaneously varied by
varying the one parameter r. For example, assigning r = 10% to all the motor operated
valves in the Emergency Core Cooling System describes the situation where all the
valves are aging with a relative increase in their failure rates of 10% per year.
116
Another important advantage in focusing on the relative aging behavior is that
component aging can then be classified into different engineering categories which
represent different degrees of aging, such as a small degree of aging, a moderate degree
of aging, or a severe degree of aging. The risk impact, eg. CDF increase, which results
from different degrees of aging can then be evaluated for given test and maintenance
practices to assess the adequacy of the practices in controlling aging effects. Table 24
gives one categorization of different degrees of aging along with the corresponding
percentage increase in the failure rate. Also given is the associated time period in which
the baseline failure rate doubles due to the aging. These values are consistent with aging
that has been observed in samples of data that have been evaluated (see for example
NUREG/CR-5510 (7)), with the significant aging corresponding to harsh environment
effects such as due to unpurified water environments or high temperatures. These
relative values appear to be applicable for a range of absolute aging rates and baseline
failure rates.
As a demonstration aging sensitivity studies, Figure 1I shows the core damage frequency
increase for a plausible range of aging behaviors up to significant aging. he plant
analyzed is the same one as analyzed in Chapter 5 for the prioritization analyses. For the
present sensitivity study, mQnthly testing (good as old testing) of all active components is
assumed. Failed components are assumed to be replaced with new components. No
burn-in problems are assumed (ie. the replaced components are good as new).
Figure 11 shows the average core damage frequency increase (per year) between
replacements versus annual percentage increase in component failure rate due to aging of
all active components. All the active components are assumed to be aging with the same
relative percentage increase in their failure rate (The baseline failure rate for each
component is that defined in the baseline PSA.). Aging of passive components and
structures is not considered in this particular sensitivity study.
The baseline core damage frequency for the plant is approximately 3x10 5 per year from
the standard PSA. This sensitivity study thus shows that aging effects are controlled over
a range of plausible aging for the given plant under the policy of monthly testing and
replacement at failure with new components. Even with a relative aging of 100%, where
all component failure rates are doubled each year, the core damage frequency increase
due to aging is still held to approximately 3xlO-5 per year which is still comparable to the
baseline core damage frequency of 3x10-5 per year. For moderate to minor aging, the
117
TABLE 24. RELATIVE AGING FAILURE RATE CATEGORIES
Category Relative Failure Rate Failure Rate Doubling

Increase (per year) Period (yrs)
Minor aging 10% 10 yrs
Moderate aging 30% 3 yrs
Significant aging 100% 1 yr
118
aging effects on the core damage frequency are controlled to be significantly less than the
baseline core damage frequency.
The aging control which is exhibited in Figure 11 is due to monthly testing with replacement at
failure. Failed components are not down for significant times until detected by a monthly test,
and failed components are replaced with new components, removing all previous aging effects.
Other policies which involve less frequent testing or which do not replace all failed components,
but only the most important contributors, could also achieve effective aging control.
Prioritization of the dominant contributors similar to that performed in Chapter 5 would identify
the components which are most important to test and replace to control aging impacts. The
dominant contributors would now need to be obtained for the range of aging rates that are
defined to be plausible.
8.3 Sensitivity Studies of the Effects of Different Test and Replacement/fRepair Policies
Sensitivity studies can also be usefully carried out to evaluate different aging management
policies for their control of aging effects on risk. Figure 12 shows the average core damage
frequency increase per year due to aging under two different surveillance test intervals. The
PSA is again the same as that used in the previous sections. Replacement at failure is again
assumed and only active component aging is considered. All active components again are
assumed to be aging with the same relative aging rate.
The monthly testing line (T = 1 month) is the same result as given in the previous section in
Figure 11. The semi-annual testing line (T = 6 months) corresponds to semi-annual testing of all
active components. With semi-annual testing the core damage frequency increase due to aging
is considerably higher and becomes larger than the baseline core damage frequency with
moderate aging (20% - 30%). Thus, monthly testing of the key contributors is important if no
additional scheduled maintenance is performed. The prioritization of contributors as done in
Chapter 5 would show those components for which it is important to control the test interval to
be 1 month.
Finally, as a sensitivity study Figure 13 shows the impact of a policy of replacement of failed
components versus a policy of repair of failed components. The replacement curve is the same
as in Figure 11 (but on a log scale). For a replacement policy again, failed components are
119
FIGURE 11. CORE DAMAGE FREQUENCY INCREASE FOR A PLAUSIBLE RANGE OF AGING
REPLACEMENT AT FAILURE: MONTHLY TESTING
3.5E-05
AC 3.OE-05
2.5E-05
2.OE-05 . . . . . . ............................................. . . . . .. .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . .
Co
1.5E-05
1.OE-05
5.OE-06 <...................... ........... I...................................... ........... ............... ..... .... .............
{ ~ ~~ I
~~~ I
O.OE+OO
0% 25% 50% 75% 100%
Relative Aging Rate (% Increase in Failure Rate Per Year)

FIGURE 12. CORE DAMAGE FREQUENCY INCREASE FOR A PLAUSIBLE RANGE OF AGING:
REPLACEMENT AT FAILURE: TWO DIFFERENT TEST INTERVALS
3.OE-04
T=1 Mo. +FT=6Mo.

AC 2.5E-04 . ..
2.OE-04 _ ... ,7
A, 1~~~~~.O E-04 . ..... . . . . . . . . . . . . . . . . ............... . ..... . . . .. . . . . . .. . . . . .. . . . . . ... ..... . . . . . . . . . .
w~~~~ .5E-
1.OE- 04 ... . . . . . . . ..
5 .OE-0 5 ... .... . .. . .. ........ ... ... ....... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
O.OE+OO
0% 25% 50% 75% 100%
FIGURE 13. CORE DAMAGE FREQUENCY INCREASE FOR A PLAUSIBLE RANGE OF AGING:
REPLACEMENT AT FAILURE VERSUS REPAIR AT FAILURE: MONTHLY TESTING
1.OE-02
_.................................................................................................................. .............................
AC
1.OE-03 . ......
0 ........ ..... . . . . .. . . . . . ........ ............ ..... ........ . ........ ...................... ...... . . . . I... . . . . . .
_ .......... ....................................................................................................................................
.... I..... ................... .......................................................................................................
1.OE-04 ..............
_ ........... ...................................................................................
1.OE-N~~.
_ . .. . . . .. . . . . . . . . . . . . . . . . . . . . .......................... ..................
................
1.OE-05 ~~~~.... .............

............ .........
............ ~~~~~~~~~~~~~~~.................. I.................
1.QE-06 -/.............
i.-::..
_
.: -................ Replacem ent - Re pr
.....................................
....................... .... .....
_....... .................................................................................. ..................................................................
1.0E-07 I I I
0% 25% 50% 75% 100%
replaced with new components. For the repair policy, it is assumed that failed
components are restored to an operational status but are not replaced with new
components. Aging effects are thus not significantly removed. The repair is thus treated
to be as good as old. The repair curve is very much higher than the replacement curve,
showing significant core damage frequency effects from aging even at minor component
aging (5% - 10%). It is thus extremely important to replace or overhaul the dominant
components instead of repairing them to control aging impacts on the core damage
frequency. Prioritization of the components would again identify those components
which need to be focused on for aging management.
123
9. CONSIDERATIONS IN USING A PSA TO EVALUATE THE RISK
EFFECTS FROM AGING OF PASSIVE COMPONENTS.
9.0 introducin
To use the PSA for evaluation of risk effects from aging of passive components, it is first
of all important that a plant specific PSA be constructed to evaluate and quantify the core
damage frequency associated with the plant's design and procedures. For the most
comprehensive PSA, all major passive components should be incorporated, otherwise it
will not be possible to explicitly evaluate the risk effects from aging of these passive
components. In this final chapter, important concepts and considerations are discussed
involving the reliability modeling of passive components and their aging effects. As part
of the risk evaluation of aging work, a separate report is planned which will discuss the
special points involved in modeling passive component aging in greater detail.
9.1 The Role and Use of PSA to Evaluate the Reliability of Passive Components and
Their Risk Impacts
PSAs have not generally included the detailed risk contributions.from passive
components and structures (e.g. pipes, vessels, containments). The passive contributions
which can be analyzed in greater detail include:
(a) Initiating events

- Primary circuit failures
- Pressure shell failures
- Steam generator tube ruptures
(b) System related contributions

- control rod failures
- pipe breaks
- vessel failures
- missile generation and impact
- pipe whip phenomena
124
(c) Containment contributions
- steel shell
- closures
Most of these contributions are presently incorporated into a PSA in only a general way
using failure rates based on generic information or expert judgement.
In particular cases there have been more formal analyses to derive a failure probability
for specific components. When more formal analyses are carried out then a probabilistic
fracture mechanics (PFM) model is often used to model the probability distributions of
various parameters including:
* flaw size and location

* failure to detect particular flaw sizes
* material property variations
* changes of material properties through life
These probability distributions are then combined to evaluate, for example, the
probability of the vessel exceeding its design criteria. The analyses are used to
determine, among other things,
* the probabilities of failure of the passive components

* the important inspection activities which have an impact on the failure
probabilities
In principle, detailed models could be developed for all major passive components and
associated failure probabilities derived.
For systems analyses performed in PSAs, as was observed from the previous
demonstrations, typically the active components (e.g., pumps, valves) are only
considered in detail. The reason is that relatively small failure rates are used for the
passive components. Also, expert judgement tends to limit the importance of passive
component contributions. As a consequence, interactions between active and passive
components (e.g., different loads as a consequence of various failure modes) are
generally incorporated in only a limited way. Aging can cause these contributions to be
more important.
125
One example of an interaction between active and passive components is the safety issue
related to pressurized thermal shock. In this event scenario, the thermal loads (stresses)
at the primary piping and reactor pressure vessel (RPV) are strongly dependent on the
function and/or malfunction of the active components in the emergency core cooling
system (ECCS) and other systems. Conversely, the failure modes of the passive
components influence the failures of the actives in this scenario. This interacting process
can be modeled in an age dependent PSA.
A PSA can thus represent a framework and an analysis tool to integrate the behavior of
active and passive components, prioritizing the importance (contribution) of each
involved component to the overall risk. The traditional deterministic safety analysis
process generates limited insights regarding the risk importances of passive components
and structures. The PSA (or PRA) can show the risk importance of different assumptions
(e.g., break location or parameters, e.g., load cycles, critical stress intensity factor) used
in these safety analyses. By incorporating probabilistic models of passive component
failures into a PSA, one can then generate answers as to not only where breaks are most
likely, but where they most effect the core damage frequency or risk. This stochastic
modeling - PSA integration combines design, manufacturing and~operating aspects and
allows a risk prioritization of the contributors and associated parameters.
9.2 Aging Component Reliability Models Required for PSA Aging Evaluations.
In aging evaluations of a plant, one is confronted with four basic phenomena causing
time dependent changes in material behavior of passive components:
- Fatigue
- Embrittlement
- Crack Growth
- Surface degradation (friction, erosion, corrosion)
Passive component modeling for PSA applications need therefore to focus on these basic
phenomena.
A passive component reliability analysis can be demonstrated where a loading stress

interacts with strength. Consider a state function S of the form,
126
- I. S = strength - stress (83)
If both the stress and the strength are random variables with defined probability
distributions, S will also be a random variable with a defined probability distribution.
When SO, the passive component is said to be in the safe state. Failure occurs when
SO (when loading and therefore stress is larger than strength). If S is normally
distributed, then the ratio of S-E(S), where E(S) is the expected value, and e(S) where
e(S) is the standard deviation is a standard normal random variable.
More generally, the probability can be computed that the structure will fail i.e., that S<O.
The state function can furthermore be extended to multiple controlling variables, with
different distributions,
S = S (X1 , x 2 , * * Xn) (84)
These types of models need to be utilized when evaluating passive component failure
probabilities (unrellabilities) to insert into the equation for the core damage frequency
increase, such as Equation (45).
9.3 The Crack Growth Phenomenon.
A probabilistic fracture mechanics crack growth analysis can be expressed in terms of a

multi-variate state function, ie. Equation (84). The result from a crack growth analysis
can be the probability versus time that the crack will propagate and cause component
failure, i.e., S-O. To carry out the reliability analysis, an initial crack size distribution
and a probability distribution for detecting a given crack size by non destructive
inspection (NDI) can be used. The effect of NDI can be explicitly incorporated using
Bayesian updating,
P(F 1) P(F)P(I/F) (85)

P(I)
where P(F/I) is the updated structural failure probability given that an inspection has
been carried out. P(F) is the structural failure probability before the inspection, P(U) is
the NDI reliability, and P(IF) is the probability of detection given the failure.
127
Instead of calculating failure probabilities, an alternative procedure is to update the crack
size only. NDI can be considered as providing new information on current crack sizes,
and can be used to constrain the updating to specific crack sizes. Bayesian approaches
can again be used to obtain associated probabilities.
9.4 The Corrosion Phenomenon
The deterioration process due to corrosion generally affects the reliability of passive
components. The corrosive effects depend on the loading process and on the
environmental conditions in the systems to which the component is exposed. Due to the
uncertainties inherent in the loading process, as well as in the manufacturing and
operating process, a variety of variables need to be considered as being random. Because
of the non-normal properties of these variables and the non-linear characteristics of the
state function, general simulation methods need to be used to estimate the failure
probability of the passive components due to corrosion effects. Normally the
deterioration process of fatigue is analyzed utilizing models that describe the crack
growth rate per load cycle.
To explicitly consider the deterioration process of corrosion, a corrosion factor C. is

often introduced. This factor depends on the loading process and the environmental
conditions to which the passive component is exposed. Based on data, the corrosion
factor C, can be approximated by a function of the stress intensity factor, the frequency
of loads and the stress ratio, and the applied loading process. Corrosion can significantly
decrease the structural reliability of passive components versus time and therefore it can
be important to consider such effects in an age dependent PSA.
9.5 Stochastic Analysis as a Complement to Deterministic Crack Growth Analysis
In the design process, and the safety assessment of operating strategies of passive
components, typically a deterministic crack growth approach (DCGA) is often used. The
DCGA provides a single value prediction for crack size at a given service time for a
single structural detail, but it does not quantity structural reliability and the importance of
different contributors. Over the past years reliable passive reactor components have been
obtained based on DCGA, good design/analysis methods and practices, quality
manufacturing, and effective inspection and quality control. However, in view of the
benefits of using a PSA, and the special issues associated with aging, a PSA approach
128
can be applied to complement existing deterministic practice (e.g. DCGA) to estimate the
failure probabilities of passive components and the importances of different passive
components as risk contributors. Passive components and interactions between passive
components and active components can then be prioritized in the same detail as active
components can. With appropriate probabilistic failure models which can include those
aging failure rate models in Chapter 2 and by expanding the PSA to include passive
component contributors, the risk effects of aging from passive components can thus be
analyzed and be prioritized in the same manner as active components are.
129
10. SUMMARY AND CONCLUSIONS
An age-dependent PSA, or an APSA, is different from a standard PSA in that an APSA

explicitly evaluates the risk effects of aging. Age dependent component failure rates are
utilized, and test and maintenance activities are explicitly analyzed for their control of
aging effects. Specific models have been presented for different aging failure rate
behaviors and for different effects of tests and maintenances in controlling aging. These
models can be incorporated into a PSA to explicitly evaluate aging effects.
Different approaches can actually be used to incorporate the aging models and to
transform a PSA to an APSA. Three basic approaches were described in detail along
with their specific features. These descriptions and the detailed models presented allow
any PSA to be transformed to an APSA.
Different applications of an APSA were described which include bottom-line

evaluations, evaluations of maintenance effectiveness, prioritization analyses, and
sensitivity and uncertainty analyses. These different applications were reviewed in light
of aging failure data which is available, test and maintenance data which is available,
PSA information which is available, and meaningful results which can be obtained.
Specific applications were then demonstrated involving prioritizations of aging

contributors, identification of risk-directed aging management strategies, evaluation of
time dependent aging effects, data uncertainty analyses, sensitivity analyses of the risk
impacts of aging, and sensitivity analyses of the effectiveness of different test and
maintenance practices in controlling the risk impacts from aging. These studies illustrate
the significant amount of useful information which is obtainable from an APSA even if
accurate aging failure rate data is not available. Finally, specific considerations were
discussed in determining the time dependent and age dependent failure probabilities for
passive components which can then be utilized in the general formulas for the aging
impacts on risk.
130
REFERENCES
1. NUREG/CR-4550, "Analysis of Core Damage Frequency from Internal Events",

Volume 1, September 1987.
2. NUREG/CR-4350, "Probabilistic Risk Assessment Course Documentation",

Volumes, 1-7, August 1985.
3. NUREG/CR-2300, "PRA Procedures Guide", Volume 1, January 1983.
4. H. Ascher and H. Feingold, Repirable Systems Reliability, Marcel Dekker, New

York, 1984.
5. MS. Abdel-Hammeed, E. Cinlar, and J. Quinn, eds, Reliability They and

Models, Academic Press, New York, 1984.
6. B.V. Gnedenko, Mathematical Methods of Reliability Theory, Academic Press,

New York, 1969.
7. NUREG/CR-55 10, "Evaluations of Core Melt Frequency Effects Due to

Component Aging and Maintenance", June 1990.
8. NUREG/CR-5248, "Prioritization of TIRGALEX-Recommended Components

for Further Aging Research", October 1988.
9. NUREG-1362, "Regulatory Analysis for Proposed Rule on Nuclear Power Plant

License Renewal", July 1990.
131
UNRELIABILiTIES AND UNAVAILABILrTIES
132
UNRELIABILrTIES AND UNAVAILABILITIES
Let X(w) be the age dependent component failure rate, where X(w) can be any form and
can include a threshold age. The component can be either an active component or
passive component. The component unreliabilityis the probability that the component
fails to operate successively for a given interval. The component unreliability can be
calculated from the following formulas depending upon the initial conditions:
= the probability of failure before age w (1)

1-exp -A(w')dw')
given the component was known to be
last up at age zero:
= the probability of failure before age w (2)

given the component was known to be
' w' last up at age wa .
When the component was last checked will depend upon the testing schedule. The above
formulas can be transformed to time dependent formulas by using the age of the
component at the given time w = w(t) .
The component unavailabilityis the probability that the component is down and is unable
to function. The component unavailability can be calculated from the following formulas
depending upon the initial conditions:
A = the probability that the component is (3)

I-exp-JA(w'
dwJ) down at age w given that the
x
component was last known to be up at
age zero:
= the probability that the component is (4)

w . r')dw'
1-exp- f)(w down at age w given the component
was last known to be up at age w. .
133
Note that formulas (3) and (4) are similar in form to formulas (1) and (2). The difference
is that operatingfailure rates are generally used in formulas (1) and (2) while standby
failure rates are used in formulas (3) and (4).
When a maintenance or restoration is performed on the component then formulas (3) and
(4) still apply with the component age measured from the last maintenance or restoration
on the component. For a restoration where the age is reset to zero then formulas (3) and
(4) apply with the age w being the age since the last restoration. For a maintenance
involving a partial restoration which resets the component age to a value w, then
formulas (3) and (4) apply with the age w being the age of the component since the last
partial restoration where the age begins at wl.
Associated time dependent formulas for the unavailability can be obtained by using the
ages that correspond to given imes. For example, if the component was renewed at to
with no subsequent checking or maintenance then the unavailability q(t) at time t is
q(t) = 1- exP(1 L(w')dw)
where the component age at time t is t-tD.
If the component was last known to be up at time % i.e. last tested at ta , then the
component unavailability q(t) is
1t-to (6)
q(t)=I-e xp- fA(w')dw'
(
ta-to
where t3 - o is the age wa at the last test or check. Similar transformations can be used
for the component unreliabilities in Equations (1) and (2).
Finally, for average component unavailabilities the above age dependent or time
dependent component unavailabilities are averaged over given time intervals. Averages
over intervals between tests give the average unavailability between tests and averages
134
over intervals between replacements give te average unavailability between
replacements. Differences from the baseline PSA unavailabilities give the increases in
unavailability (or unreliability) due to aging, either for time dependent or for averaged
results.
135
NRC FORM 335 U.S. NUCLEAR REGULATORY COMMISSION 1. REPORT NUMBER
(2-89) EAuunWd by NRC. Add Vol.Sp., Re.
NRCM 1102, and Addendum Numbers If any.
3201,3202 BIBLIOGRAPHIC DATA SHEET
(See isttons on the reverse) NUREG/CR-5587
2. TITLE AND SUBTITLE SAIC-92/1137
Approaches for Age-Dependent Probabilistic Safety
Assessments with Emphasis on Prioritization and . DATE REPORT PUBLISHED
Sensitivity Studies MONt1

August 1992
4. FIN OR GRANT NUMBER
L1072
5. AUTHOR(S) 6. TYPE OF REPORT
Technical
7. PERIOD COVERED f(ic/lswve Dites)
W.E. Vesely
S. PERFORMING ORGANIZATION - NAME AND ADDRESS If NRC. provide Division, Office or Reon, US Nuckar RegulatorY miion anmili Art; If contractor provide
um end meilivaddss.)

655 Metro Place South, Suite 745
Dublin, OH 43017
9. SPONSORING ORGANIZATION -NAME AND ADDRESS f NRC, type 'Sane as aove" If contracto, provke NRC Divion, Offkc or Region, U.S Nucear Reulatory Commission,
anedfilng addreVsJ
Division of Engineering
Office of Nuclear Regualtory Research
U.S. Nuclear Regulatory Comission
Washington, DC 20555
10. SUPPLEMENTARY NOTES
11. ABSTRACT 2W wory or a
Approaches are described for incorporating component aging reliability models into
a probabilistic safety assessment (PSA), or probabilistic risk assessment (PRA), of
a nuclear power plant. These approaches and procedures are described from a
technical standpoint and are not to be interpreted as having any regulatory
implications. Component aging failure rate models and test and maintenance aging
control models are presented for utilization. Different approaches for carrying
out the aging evaluations are given. Demonstrations are given involving prioritizing
aging contributors, evaluating maintenance effectiveness, carrying out time dependent
evaluations, and carrying out uncertainty and sensitivity analyses of aging effects.
12. KEY WORDS/DESCRIPTORS (ist words or phrseth will asit weaheinocaetherort.) 13. AVAILABILITY STATEMENT
Probabilistic Safety Assessment Unlimited

Probabilistic Risk Assessment 14. SECURITY CLASSIFICATION
Aging Evaluations (This Pe)

Component Reliability Mbdels Unclassified
This Report)
Unclassified
15. NUMBER OF PAGES
1 . PRICE
NRC FORM 335 2491

Federal Recycling Program
UNITED STATES SPECIAL FOURTH-CLASS RATE
NUCLEAR REGULATORY COMMISSION POSTAGE AND FEES PAID
WASHINGTON, D.C. 20555-0001 USNRC
PERMIT NO. 0-67
OFFICIAL BUSINESS USNRC

PENALTY FOR PRIVATE USE, $300 r4LS

NUREG-CR-5587 Approaches For Age-Dependent PSAs Etc

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

NUREG-CR-5587 Approaches For Age-Dependent PSAs Etc

Diunggah oleh

Hak Cipta:

Format Tersedia

NUREG/CR-5587

Approaches for Age-Dependent

Science Applications International Corporation

3. The National Technical Information Service, Springfield, VA 22161

Approaches for Age-Dependent

Manuscript Completed: May 1992

Science Applications International Corporation

1. THE AGE-DEPENDENT PSA VERSUS THE STANDARD PSA............. 1

2. COMPONENT RELIABILITY MODELS USED IN AN

2.0 Introduction. ........ ,......... 17

3. APPROACHES FOR TRANSFORMING A PSA INTO AN AGE-

3.0 Introduction ................................................ 45

4. APPLICATIONS OF AN AGE-DEPENDENT PSA ................................. 58

4.0 Introduction ............................................... 58

5. PRIORITIZATIONS OF AGING CONTRIBUTORS ................................. 67

5.0 Introduction ....... 67

6. IDENTIFICATION OF RISK-DIRECTED AGING MANAGEMENT 94

6.0 Introduction .94

7. TIME DEPENDENT AGING EVALUATIONS ....................................... 104

7.0 Introduction ............................................. 104

8. SENSITIVITY AND UNCERTAINTY EVALUATIONS .......................... 111

8.0 Introduction . ............................................. l

9. CONSIDERATIONS IN USING A PSA TO EVALUATE THE RISK

9.0 Introduction .................................................. 124

10. SUMMARY AND CONCLUSIONS ........................ 130

REFERENCES ........................................................ ............................................... 131

FIGURE 1. BASIC STEPS IN A PSA................................................................ 2

FIGURE 2. THE BATHTUB CURVE OF FAILURE RATE VERSUS AGE.... 5

FIGURE 3. INCREASE IN CORE DAMAGE FREQUENCY FROM AGING

FIGURE 4. TIME DEPENDENT CORE DAMAGE FREQUENCY

FIGURE 5. SENSITIVITY ANALYSIS EVALUATING THE AGING

FIGURE 6. QUESTIONS TO BE ANSWERED IN SETTING UP AN

FIGURE 7. CORE DAMAGE FREQUENCY INCREASE DUE TO AGING:

FIGURE 8. RELATIVE CORE DAMAGE FREQUENCY INCREASE DUE

FIGURE 9. TOM DEPENDENT CORE DAMAGE FREQUENCY

FIGURE 10. UNCERTAINTY ANALYSIS OF AGING EFFECTS: PLANT A 114

FIGURE 11. CORE DAMAGE FREQUENCY INCREASE FOR A

FIGURE 12. CORE DAMAGE FREQUENCY INCREASE FOR A

FIGURE 13. CORE DAMAGE FREQUENCY INCREASE FOR A

TABLE 1. THE DIFFERENCES IN COMPONENT RELIABILITY

TABLE 2. PRIORITIZED COMPONENT AGING CONTRIBUTORS......... 10

TABLE 3. ISSUES ASSOCIATED WITH AN APSA.................................... 16

TABLE 4. QUESTIONS AND ALTERNATIVE ANSWERS TO

TABLE 5. THE SUCCESSIVE STEPWISE APPROACH FOR PSA

TABLE 6. THE SUBSTITUTION APPROACH FOR PSA AGING

TABLE 7. THE RISK IMPORTANCE APPROACH FOR PSA AGING 57

TABLE 8. TIRGALEX AGING RATES USED FOR ACTIVE

TABLE 9. LOW SENSITIVITY AGING RATES USED FOR

TABLE 10. HIGH SENSUVY AGING RATES USED FOR

TABLE 11. TIRGALEX SURVEILLANCE TEST EF ICIENCIES ............... 76

TABLE 12A. CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

TABLE 12B. CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

TABLE 13A: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

TABLE 13B: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

TABLE 14A: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

TABLE 14B: CORE DAMAGE FREQUENCY INCREASE FROM ACTIVE

TABLE 15. COMPONENT IDENTIFIERS ..................... - . 87

TABLE 16. COMPONENTS PRIORITIZED BY ORDER OF MAGNITUDE

TABLE 17. COMPONENTS PRIORITIZED BY ORDER OF MAGNITUDE

USNRC ORDER NUMB: A65617-00006

DOCUMENT NO. QTY a COST TOTAL COST COMMENTS

TOTAL DOCUMENTS ORDERED 5

TABLE 18. COMPONENTS PRIORITIZED BY ORDER OF MAGNITUDE